Simulated Ground-Water Flow in the Hueco Bolson, an Alluvial-Basin Aquifer System near El Paso, Texas
U.S. DEPARTMENT OF THE INTERIOR U.S. GEOLOGICAL SURVEY
Water-Resources Investigations Report 02-4108
Prepared in cooperation with EL PASO WATER UTILITIES and the U.S. ARMY- FORT BLISS Simulated Ground-Water Flow in the Hueco Bolson, an Alluvial-Basin Aquifer System near El Paso, Texas
By Charles E. Heywood and Richard M. Yager
U.S. GEOLOGICAL SURVEY
Water-Resources Investigations Report 02-4108
Prepared in cooperation with EL PASO WATER UTILITIES and the U.S. ARMY - FORT BLISS
Albuquerque, New Mexico 2003 U.S. DEPARTMENT OF THE INTERIOR GALE A. NORTON, Secretary
U.S. GEOLOGICAL SURVEY Charles G. Groat, Director
The use of firm, trade, and brand names in this report is for identification purposes only and does not constitute endorsement by the U.S. Geological Survey.
For additional information write to: Copies of this report can be purchased from:
District Chief U.S. Geological Survey U.S. Geological Survey Information Services Water Resources Division Box 25286 5338 Montgomery Blvd. NE, Suite 400 Denver, CO 80225-0286 Albuquerque, NM 87109-1311
Information regarding research and data-collection programs of the U.S. Geological Survey is available on the Internet via the World Wide Web. You may connect to the home page for the New Mexico District Office using the URL http://nm.water.usgs.gov. CONTENTS Page Abstract...... 1 Introduction ...... 1 Purpose and scope ...... 2 Previous investigations ...... 2 Hydrogeology of the Hueco Bolson ...... 2 Extent and thickness of hydrogeologic facies ...... 5 Hydraulic conductivities...... 5 Ground-water levels...... 5 Rio Grande Valley surface-water system...... 8 Agricultural canals and drains ...... 12 Recharge ...... 12 Mountain-front recharge...... 12 Underflow from Tularosa and Mesilla Basins ...... 13 Human-induced recharge...... 13 Discharge ...... 13 Ground-water withdrawals ...... 13 Evapotranspiration...... 13 Seepage to Rio Grande and agricultural drains ...... 13 Land subsidence ...... 14 Steady-state and transient ground-water flow model...... 14 Numerical method ...... 15 Aquifer dewatering ...... 15 Multi-aquifer wells ...... 15 Spatial discretization...... 18 Vertical datums ...... 18 Temporal discretization ...... 18 Boundary conditions...... 19 Specified-flow boundaries ...... 19 Head-dependent flow boundaries ...... 19 Streamflow routing ...... 20 Faults...... 23 Storage properties...... 23 Model calibration...... 23 Hydraulic-head measurements ...... 23 Flow measurements ...... 23 Parameter estimation ...... 23 Model evaluation and simulation results ...... 26 Estimates of aquifer properties...... 26 Parameter sensitivities ...... 26 Water-level drawdowns...... 29 Ground-water budget...... 29 Seepage from the Rio Grande...... 29 Summary and conclusions...... 51 Selected references ...... 53 Appendix 1: Modifications to MODFLOWP ...... 56 Appendix 2: Modifications to MODFLOW ...... 63 Appendix 3: Multi-aquifer well package...... 68
iii FIGURES Page 1. Shaded-relief map showing principal physiographic features and thickness of alluvial deposits in the Hueco Bolson...... 3 2. Shaded-relief map showing ground-water flow-model domain, location of wells, and trace of geologic sections shown in figure 3...... 4 3. Generalized geologic sections A-A′ along the western margin of the Hueco Bolson and B-B′ along an east-west transect north of El Paso, Texas ...... 6 4. Cutaway perspective view showing generalized horizontal hydraulic conductivity of hydrogeologic facies...... 7 5. Graph showing ground-water withdrawals from the Hueco Bolson, 1903-96...... 8 6. Hydrographs showing water-level declines in selected wells in and near El Paso and Ciudad Juarez, 1935-96.... 9 7. Graphs showing increases in chloride concentration in selected wells, 1951-96...... 10 8. Schematic showing stream segments used for routing surface-water flows...... 11 9. Schematic diagrams showing modifications to MODFLOWP and MODFLOW: A. Computation of head in partially dewatered, multilayer observation wells ...... 16 B. Stream leakage through dewatered model cell ...... 16 C. Distribution of hydraulic head and flow near multilayer pumped well ...... 17 10. Graphs showing mean annual and mean monthly flows in Rio Grande and diversions, 1903-96 ...... 22 11. Map showing locations of head measurements, stream segments, agricultural drains, and model faults...... 24 12. Graph showing number of annual head measurements in wells included in the nonlinear regression used in the ground-water flow model...... 25 13. Graphs showing simulated and measured heads and weighted residuals computed in transient-state simulation of the Hueco Bolson...... 27 14. Maps showing simulated water table in shallow aquifer (model layers 1 and 2) in: A. 1902 (steady-state conditions)...... 30 B. 1958 ...... 31 C. 1973 ...... 32 D. 1980 ...... 33 15. Maps showing simulated potentiometric surface in model layer 5 in: A. 1958 ...... 34 B. 1973 ...... 35 C. 1980 ...... 36 D. 1996 ...... 37 16. Perspective views showing water-level drawdown in the Hueco Bolson aquifer in: A. 1973 ...... 38 B. 1996 ...... 39 17. Hydrographs showing measured and simulated water levels from 1930 through 1995 in selected wells in the: A. Mesa-Nevins well field...... 40 B. Airport well field ...... 42 C. Cielo Vista and Town well fields ...... 43 D. Water-plant well field ...... 44 E. Juarez well field...... 45 F. East observation well field...... 47 G. Lower valley and north observation well fields...... 48 18. Graphs showing simulated annual inflows and outflows to Hueco Bolson aquifer: A. Ground-water withdrawals...... 49 B. Storage...... 49 C. Stream leakage...... 49 D. Agricultural drains and evapotranspiration ...... 49 19. Graphs showing Rio Grande seepage between Chamizal zone and Riverside Dam ...... 50 20. Graph showing simulated and measured flows at Riverside Dam ...... 52
iv TABLES 1. Measured and simulated flow loss from unlined section of the Rio Grande above Riverside Dam and Franklin Canal...... 12 2. Conversions between vertical datums in the El Paso, Texas, area...... 18 3. Optimum parameter values estimated for Hueco Bolson aquifer, through nonlinear regression in transient- state simulation, and their approximate 95-percent confidence intervals ...... 28
CONVERSION FACTORS, ABBREVIATIONS, AND DATUMS
Multiply By To obtain
centimeter (cm) 0.06102 inch (in.) millimeter 0.03937 inch (in.) meter (m) 3.281 foot (ft) kilometer (km) 0.6214 mile (mi) square kilometer (km2) 0.3861 square mile (mi2) square kilometer (km2) 247.1 acre cubic meter (m3) 0.0008107 acre-foot (acre-ft) cubic meter (m3) 35.31 cubic foot (ft3) cubic meter (m3) 264.2 gallon (gal) cubic meter per day (m3/d) 0.0004087 cubic foot per second (ft3/s) cubic meter per day (m3/d) 0.2961 acre-foot per year (acre-ft/yr cubic meter per day (m3/d) 96,489 gallons per year (gal/yr)
acre-foot (acre-ft) 1,233 cubic meter (m3)
Altitude, as used in this report, refers to distance above sea level.
Sea level: In this report, "sea level" refers to the National Geodetic Vertical Datum of 1929—a geodetic datum derived from a general adjustment of the first-order level nets of the United States and Canada, formerly called Sea Level Datum of 1929.
v SIMULATED GROUND-WATER FLOW IN THE HUECO BOLSON, AN ALLUVIAL-BASIN AQUIFER SYSTEM NEAR EL PASO, TEXAS By Charles E. Heywood and Richard M. Yager
ABSTRACT The model input was generated from geographic information system databases, which facilitated rapid The neighboring cities of El Paso, Texas, and model construction and enabled testing of several Ciudad Juarez, Chihuahua, Mexico, have historically conceptualizations of hydrogeologic facies boundaries. relied on ground-water withdrawals from the Hueco Specific yield of unconfined layers and hydraulic Bolson, an alluvial-aquifer system, to supply water to conductance of Quaternary faults in the fluvial facies their growing populations. By 1996, ground-water were the most sensitive model parameters, suggesting drawdown exceeded 60 meters in some areas under that ground-water flow is impeded across the fault Ciudad Juarez and El Paso. planes. A simulation of steady-state and transient ground-water flow in the Hueco Bolson in westernmost Texas, south-central New Mexico, and northern INTRODUCTION Chihuahua, Mexico, was developed using MODFLOW-96. The model is needed by El Paso The neighboring cities of El Paso, Texas, and Water Utilities to evaluate strategies for obtaining the Ciudad Juarez, Chihuahua, Mexico, have historically most beneficial use of the Hueco Bolson aquifer relied on ground-water withdrawals from the Hueco system. The transient simulation represents a period of Bolson, an alluvial-aquifer system, to supply water to 100 years beginning in 1903 and ending in 2002. The their growing populations. At the end of the 20th period 1903 through 1968 was represented with 66 century, about 680,000 people lived in the El Paso annual stress periods, and the period 1969 through metropolitan area (U.S. Department of Commerce, 2002 was represented with 408 monthly stress periods. 2002) and the total population (including Ciudad The ground-water flow model was calibrated Juarez) was about 2 million. The term “bolson,” using MODFLOWP and UCODE. Parameter values literally meaning “handbag” or “purse” in Spanish, is representing aquifer properties and boundary local terminology that may be considered synonymous conditions were adjusted through nonlinear regression with “basin.” The Hueco Bolson is considered the in a transient-state simulation with 96 annual time steps southern portion of the Tularosa-Hueco Basin. The to produce a model that approximated (1) 4,352 water term “Tularosa Basin” is used for the northern portion, levels measured in 292 wells from 1912 to 1995, (2) which lies entirely in the State of New Mexico and is three seepage-loss rates from a reach of the Rio Grande not modeled in this study. during periods from 1979 to 1981, (3) three seepage- In the United States, diversions from the Rio loss rates from a reach of the Franklin Canal during Grande and ground-water withdrawals from the periods from 1990 to 1992, and (4) 24 seepage rates Mesilla Basin (west of the study area) also supply the into irrigation drains from 1961 to 1983. Once a freshwater demands of the military, industries, and calibrated model was obtained with MODFLOWP and public in the El Paso area. In Mexico, diversions from UCODE, the optimal parameter set was used to create the Rio Grande are used for agriculture; water needed an equivalent MODFLOW-96 simulation with monthly by Ciudad Juarez is supplied solely by extraction from temporal discretization to improve computations of the Hueco Bolson. By 1996, ground-water drawdown seepage from the Rio Grande and to define the flow exceeded 60 m in some areas under Ciudad Juarez and field for a chloride-transport simulation. El Paso. Fresh ground water stored in the aquifer Model boundary conditions were modified at system beneath these cities is bordered by regions of appropriate times during the simulation to represent brackish to saline ground water. As water levels in the changes in well pumpage, drainage of agricultural freshwater portions of the aquifer declined, intrusion of fields, and channel modifications of the Rio Grande. the surrounding brackish water degraded water quality
1 in public supply wells, which sometimes required well Previous Investigations abandonment. Prior to human intervention, infiltration of water Sayre and Livingston (1945) first provided a from the Rio Grande was the dominant mechanism of comprehensive overview of ground-water resources in aquifer-system recharge in the El Paso area. During the the El Paso area. Several ground-water flow models 20th century, numerous diversions from the Rio have been developed to investigate the effects of Grande affected the distribution of this potential pumping on water levels and salinity in the Hueco recharge water. A section of the Rio Grande between Bolson. Leggat and Davis (1966) constructed an downtown El Paso and Ciudad Juarez was converted to electric analog model to predict ground-water a lined canal in 1968, which prevented subsequent drawdowns through 1990 resulting from proposed recharge from the river to the ground-water system ground-water withdrawals. A two-layer transient model by Meyer (1976) represented freshwater with a along that section. This decreased recharge dissolved-solids concentration less than 1,000 exacerbated the effect of increased ground-water milligrams per liter in both alluvial and bolson withdrawals, increasing the rate of ground-water-level deposits. The model was used to estimate the total declines (Land and Armstrong, 1985). volume of freshwater in storage and to simulate water- level declines resulting from planned ground-water Purpose and Scope withdrawals from 1973 to 1991. Lee Wilson and Associates (1985a,b; 1991) This report describes the hydrogeology of the developed a four-layer model using MODFLOW in Hueco Bolson and documents a transient ground-water which layer thickness corresponded to the presumed flow model of the Hueco Bolson. The model, thickness of water-quality zones. Kernodle (1992) used developed in cooperation with El Paso Water Utilities the Lee Wilson and Associates (1985a,b) model to (EPWU) and the U.S. Army at Fort Bliss, is needed by estimate additional elastic aquifer compaction that the EPWU to evaluate strategies for obtaining the most might result from diverting flow in a segment of the Rio beneficial use of the Hueco Bolson aquifer system. Grande into an extension of the American Canal. Included in the report are a (1) description of the Groschen (1994) developed a four-layer model using MODFLOW and HST3D (Kipp, 1987) to simulate the hydrogeologic features relevant to the numerical movement of saline water in response to ground-water simulation, (2) summary of the computer codes used withdrawal and concluded that increased salinity in for the simulations, (3) description of the model- wells screened in the bolson deposits was caused by calibration procedure, and (4) discussion of the results leakage from the overlying alluvial aquifer. of the steady-state and transient simulations, including estimation of seepage loss with a specified design flow in the Rio Grande and no diversion to the American Canal Extension (ACE). Modifications to HYDROGEOLOGY OF THE HUECO MODFLOWP and MODFLOW are presented in two BOLSON appendixes. The Hueco Bolson is a fault-bounded structural 2 The active model area encompasses 5,303 km depression associated with the Rio Grande Rift (fig. 1). 2 (2,408 mi ) in Mexico and the United States, extending At the inception of Rio Grande rifting about 26 million from 37 km north of the New Mexico-Texas State years ago (Chapin and Seager, 1975), normal faults boundary to a point on the Rio Grande 3 km south of accommodated regional extension, resulting in Fort Hancock, Texas, 79 km southeast of El Paso. downdropped structural grabens. Igneous rocks of Model boundaries represent the perimeter of Hueco Precambrian age and sedimentary rocks of Paleozoic Bolson deposits throughout most of the modeled area and Mesozoic age surround and underlie the Hueco and coincide with the Franklin and Organ Mountains to Bolson. Unconsolidated to poorly consolidated the west and the Hueco Mountains to the east in the deposits of Tertiary and Quaternary age consisting United States, and the Sierra Juarez, Sierra El Presidio, primarily of gravel, sand, silt, and clay have filled the and Sierra Guadalupe to the west and south in Mexico basin. These deposits compose the alluvial-aquifer (figs. 1 and 2). system known as the Hueco Bolson.
2 106˚30´ 106˚15´
n New iftMexico asi B 32˚ Mountains sa o 15´ Organ lar White u T Sands Missile Rio Grande R Range Hueco
300 Bolson Texas Mexico 106˚00´
1200
900
NEW MEXICO 32˚ 60 00´ 1800 900 0 TEXAS H
2700 ueco Franklin
Mountains Mountains
NM EXPLANATION El Paso 300 Line of equal thickness of alluvium. Interval 300 meters
31˚ The 45´ Narrows Sierra R Ciudad i u o J ar Juarez e z Grande
600 MEXICO
31˚ 30´
2400
Sierra 900
E 1200 1800 l Presidi Sierra G uadalupe o
Shaded-relief map from U.S. Geological Survey 0 10 20 KILOMETERS 30-meter Digital Elevation Models 01020 MILES Figure 1. Principal physiographic features and thickness of alluvial deposits in the Hueco Bolson.
3 106˚30´ 106˚15´ 106˚00´
32˚ 15´ A´
32˚ 00´
B´ El Paso B
31˚
45´ A cequia
M Fr adre anklin
C A anal
Riverside
Canal
31˚ 30´ Rio
EXPLANATION G rande Trace of A A´ geologic section
Model boundary
Rivers and canals
Wells
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 2. Ground-water flow-model domain, location of wells, and trace of geologic sections shown in figure 3.
4 Extent and Thickness of Hydrogeologic gradational changes in hydraulic conductivity at the Facies scale of the ground-water flow model.
The alluvial deposits that make up the Hueco Bolson can be classified into four hydrogeologic facies Hydraulic Conductivities (collectively, the bolson-fill facies) on the basis of their depositional processes and resulting sedimentary EPWU has conducted aquifer pumping tests in structures: 85 production wells. Production wells are generally (1) Fluvial facies. From 3.8 million to 0.67 installed in known high-permeability areas, such as the million years ago, the ancestral Rio Grande meandered alluvial-fan and fluvial facies in the Hueco Bolson. The south along the east side of the Franklin Mountains, depositing a thick sequence of fluvial sediments average horizontal hydraulic conductivity estimated consisting of fine- to coarse-grained channel sand from these tests is 10 m/d, with a standard deviation of interbedded with silt and clay overbank deposits. The 7 m/d. The minimum and maximum measured predominant geologic formation in this facies is the horizontal hydraulic conductivities from these tests Camp Rice Formation (Strain, 1969) of Tertiary and were 1 and 50 m/d, respectively. Quaternary age. Electric logs of 101 wells in the El Laboratory measurements of permeability and Paso area indicate that the fraction of clay interbeds compressibility were made on clay core samples from within the freshwater portion of this facies is approximately one-third. five different depth intervals at a site near the Rio (2) Alluvial-fan facies. Alluvial fans originating Grande (Harold Olsen, Colorado School of Mines, from the present-day Organ and Franklin Mountains written commun., 1995). Vertical hydraulic and Sierra Juarez consist of poorly sorted gravel and conductivities of the undisturbed samples ranged from coarse- to fine-grained sand. Deposits of this facies 6 x 10-3 to 2 x 10-2 m/d. interfinger with the fluvial deposits of the Rio Grande. (3) Lacustrine-playa facies. Thick deposits of clay and silt exist in the east and southeast parts of the Ground-Water Levels Hueco Bolson and at depth beneath the fluvial and alluvial-fan facies. These fine-grained sediments were Extensive ground-water pumping from the deposited in a low-energy environment, possibly a lake Hueco Bolson since the 1940’s has resulted in cones of of mid-Cenozoic age that formed a terminal depocenter depression in the water table under El Paso and Ciudad for the ancestral Rio Grande (Strain, 1969). The Juarez. The observed drawdown cones generally predominant geologic formation of this facies is the Tertiary Fort Hancock Formation. correspond with the extent of the major production (4) Recent alluvial facies. Deposition of the wells shown in figure 2. Total ground-water pumpage fluvial, alluvial-fan, and lacustrine-playa facies and from 1903 through 1996 from the Hueco Bolson by the subsequent erosion resulted in formation of the United States and Mexico is shown in figure 5. Ground- topographic mesa that today is east and north of El Paso water withdrawals started to increase substantially (Langford, 2001). About 0.67 million years ago, the during the 1950’s drought years and have increased in Rio Grande breached “The Narrows” (fig. 1), the gap Mexico since the early 1970’s. Hydrographs of water separating the present-day Franklin Mountains from the Sierra Juarez. The Rio Grande eroded the levels measured in the United States and Mexico from topographic mesa, forming the present-day Rio Grande 1935 to 1996 illustrate that ground-water levels have Valley. Approximately 30 to 60 m of late Pleistocene to declined concurrently with pumpage. Examples of recent sediments associated with the modern Rio declining water levels in selected wells in the United Grande have been deposited in the Rio Grande Valley. States are shown in figure 6. Water quality, particularly The cumulative thickness of these alluvial in regard to chloride concentration, has degraded in deposits in the Hueco Bolson is mapped in figure 1. some areas of ground-water development from 1951 to The distribution of these facies is illustrated in the 1996 (fig. 7), such as in well 86 near the El Paso airport generalized geologic sections in figure 3 and in a cutaway perspective view in figure 4. The horizontal and well 148 in the Rio Grande Valley. In these wells, boundaries between the fluvial and lacustrine-playa ground-water pumpage may have induced intrusion facies are believed to interfinger, resulting in from adjacent or overlying regions of brackish water.
5 SOUTH NORTH A Bend in Bend in A' section section
Land surface
ande 1,200 1903 water table 1995 water table Rio Gr Model layer 1 2 3 4 5 Layers 1,000 6 1 to 9 7 8 9 , in meters 10
800 Altitude Layer
k 10 Bedrock Bedroc 600 0 10,000 20,000 30,000 40,000 50,000 60,000 Distance, in meters Vertical exaggeration = 33.9:1
WEST Bend in EAST B section B' Land surface
1,200 1903 water table 1995 water table Model layer 1 2 3 4 1,000 5 Layers 6 1 to 9 7 8
, in meters 9 10 800 Altitude
Layer 10
Bedrock Bedrock 600 0 5,000 10,000 15,000 20,000 25,000 Distance, in meters Datum is sea level Vertical exaggeration = 16.6:1
EXPLANATION Alluvial-fan facies Lacustrine-playa facies
Fluvial facies Recent alluvial facies
Figure 3. Generalized geologic sections A-A´ along the western margin of the Hueco Bolson and B-B´ along an east-west transect north of El Paso, Texas (trace of sections shown in figure 2).
6 NORTH
Horizontal hydraulic conductivity, in meters per day 6.8
5.4
3.9
2.4
0.92
Figure 4. Cutaway perspective view showing generalized horizontal hydraulic conductivity of hydrogeologic facies (logarithmic color scale bar). Area of block is identical to that shown in figure 2.
7 250
Total pumpage in model Reported by Mexico 200
150
, in millions of cubic meters 100 Pumpage
50
0 1903 1906 1909 1912 1915 1918 1921 1924 1927 1930 1933 1936 1939 1942 1945 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996
Figure 5. Ground-water withdrawals from the Hueco Bolson, 1903-96.
Rio Grande Valley Surface-Water System decline near the American Canal accelerated following this diversion (Land and Armstrong, 1985), suggesting The Rio Grande is hydraulically connected to the that decreased local aquifer recharge from the Rio Hueco Bolson aquifer system. Locally, water from the Grande exacerbated these ground-water drawdowns. river seeps into the shallow part of the aquifer system The ACE, completed in 1999, conveys water that in the Rio Grande Valley. Much of this water may be formerly flowed in a 15-km (9.3-mi) reach of the Rio transpired by agricultural crops and natural vegetation Grande channel between the Chamizal zone and in the valley. Seepage from the Rio Grande between El Riverside Dam (fig. 8). The canal was constructed, in Paso and Ciudad Juarez decreased substantially after part, to salvage water “lost” to seepage through the December 1968, when flow was diverted into the riverbed to the underlying shallow aquifer. This concrete-lined American Canal upstream from the seepage was estimated for 1981 through 1983 by the Chamizal zone. International Boundary and Water Commission The Chamizal (named after the desert shrub (IBWC) (Land and Armstrong, 1985; White and Chamiza) is a zone adjacent to the Rio Grande along others, 1997). These seepage losses are summarized in the border between El Paso and Ciudad Juarez. table 1; the average loss for the 3 years was 1.03 x 105 Originally part of Mexico, the Chamizal became part of m3/d (83.3 acre-ft/d). The magnitude of this seepage the United States when the border-defining Rio Grande varied as the hydraulic gradient changed between the changed course following a flood in 1864. In 1968, the Rio Grande and underlying aquifer. Ground-water Chamizal returned to Mexican ownership when the levels generally declined under this reach through the border was redefined along the newly constructed 1980’s and 1990’s; thus, seepage from the Rio Grande American Canal. At that time, the reach of the natural probably increased during this period. The magnitude river channel south of the Chamizal was abandoned, of this seepage (and by implication, the quantity of and flow in the Rio Grande was diverted into the water “salvaged”) is of interest to water management American Canal. The rate of ground-water-level parties in the El Paso area.
8 1,115 JL-49-13-610 (well 86) 1,105
1,095
1,085
1,075
1,115 JL-49-13-804 (well 104)
1,105
1,095
1,085
1,075
1,115 JMAS 28 (well 207) W a t e r l v , i n m s b o 1,105
1,095
1,085
1,075 1935 1945 1955 1965 1975 1985
Figure 6. Water-level declines in selected wells in and near El Paso and Ciudad Juarez, 1935-96 (location of wells shown in figure 11).
9 1,100 JL-49-13-610 (well 86) 1,000
900
800
700
600
500
400
300
200
100
0 1950 1960 1970 1980 1990
1,100
1,000 JL-49-21-301 (well 148) 900
800
700
600
500
C h l o r i d e c o n c e n t r a t i o n , i n p a r t s p e r m i l l i o n C h l o r i d e c n t a , p s m 400
300
200
100
0 1950 1960 1970 1980 1990
Figure 7. Increases in chloride concentration in selected wells, 1951-96 (location of wells shown in figure 11).
10 Canal
Riverside n io s n e t Dam x
e Riverside
l
a
n a C an ric e Ascarate wasteway nde m ra A G io flows. surface-water R g
Franklin Canal Surface-water accounting components ments used for routin ments used for g
Treatment Plant Treatment Canal one Haskell Waste Water stream se
Chamizal z g American Plant Canal Water Treatment Leon wasteway adre Schematic showin ure 8. Basin Acequia M g Settling wasteway Fi Flow measuring station Extensometer site Direction of flow Dam Dam Mexico American International NOT TO SCALE
11 Table 1. Measured and simulated flow loss from unlined Recharge section of the Rio Grande above Riverside Dam and Franklin Canal Precipitation over the Hueco Bolson is highly [IBWC, International Boundary and Water Commission; variable, both spatially and temporally. Mean annual BOR, Bureau of Reclamation; USGS, U.S. Geological precipitation over the Hueco Bolson is less than 25 cm Survey; m3/d, cubic meters per day] (10 in.), most of which falls during the summer months. Sparse rainfall over the basin floor outside the Rio Measured Simulated Measurement flow loss flow loss Grande Valley probably evaporates or transpires from the vadose zone before it can infiltrate to water-table Rio Grande 1981 8.4 x 104 m3/d 9.64 x 104 m3/d (IBWC) depths and recharge the aquifer system. In the Rio Grande Valley, where the water table can be within Rio Grande 1982 1.06 x 105 m3/d 9.82 x 104 m3/d (IBWC) several meters of land surface, precipitation or applied irrigation water has a better chance of infiltrating to the Rio Grande 1983 1.18 x 105 m3/d 1.02 x 105 m3/d (IBWC) water table. Concentrated surface flows in arroyos below mountain canyons may infiltrate sufficiently to Franklin Canal 1984 1.05 x 104 m3/d 9.81 x 103 m3/d (BOR) penetrate the vadose zone to the water table.
Franklin Canal 1990 5.06 x 104 m3/d 7.79 x 103 m3/d Ground-water age, defined as the time since (USGS) water was in contact with the atmosphere, was Franklin Canal 1991 5.42 x 104 m3/d 1.23 x 104 m3/d estimated using carbon-14 age-dating techniques at (USGS) eight different sites in the Hueco Bolson (Anderholm Franklin Canal 1992 5.33 x 104 m3/d 1.32 x 104 m3/d and Heywood, 2003). The calculated age of water in (USGS) these samples ranges from 12,100 to 25,500 years old. The dates indicate that this water advected from recharge locations for some distance through the Agricultural Canals and Drains saturated zone. Because the American Canal and its extension are concrete lined, they do not substantially interact Mountain-Front Recharge with the shallow ground-water system. The Franklin By assuming that 25 percent of precipitation that Canal in the United States and the Acequia Madre in falls in the catchments of the Organ and Franklin Mexico are major unlined irrigation-supply canals. Mountains may bypass caliche layers at mountain These canals supply water through numerous fronts, Sayre and Livingston (1945) estimated that subsidiary unlined irrigation canals to agricultural recharge to the bolson-fill aquifer from these areas may fields in the lower Rio Grande Valley. be as much as 50,000 m3/d (15,000 acre-ft/yr). Meyer To control shallow ground-water levels and (1976) estimated total recharge from the Organ and prevent soil salinization, agricultural drains were Franklin Mountains, the Sierra Juarez, and underflow installed in the Rio Grande Valley near El Paso from the Tularosa Basin to be about 19,000 m3/d beginning in the 1930’s. These drains originally were (5,640 acre-ft/yr). Wilkins (1998) estimated recharge designed and constructed with constant gradient; the along the western boundary of the Tularosa-Hueco Bureau of Reclamation (BOR) surveyed these drain- 3 3 bed altitudes in the early 1960’s. Since the 1960’s, Basin to be about 245 m /d/km (0.161 ft /s/mi). For the some drains have been destroyed by development, such boundary length corresponding to the Organ and as the construction of the ACE. From 1960 through Franklin Mountains and Sierra Juarez in this study 2000, the original gradient of many drains was not (approximately 80 km), this recharge estimate equates 3 maintained (Al Blair, AWBlair Engineering, oral to about 19,600 m /d. Waltemeyer (2001) used the commun., June 2001). Infilling from windblown sand basin-climatic characteristics method to estimate and lateral collapse increased drain-bed altitudes in streamflow available for potential recharge at the segments of the lower valley drains. This may have mouth of two canyons at the base of the Organ decreased drainage efficiency, leading to the observed Mountains, which are in the area of this model. Total higher shallow ground-water levels and increased soil mean annual streamflow from the Oak and Soledad salinity in some lower valley agricultural fields. Canyon drainages was estimated to be 2,300 m3/d;
12 some fraction of this flow may contribute to aquifer Discharge recharge. Discharge from the deeper portions of the Hueco Underflow from Tularosa and Mesilla Basins Bolson under valley and mesa areas is principally from ground-water pumping. Ground-water discharge Precipitation falling on the Sacramento and San mechanisms from recent alluvial deposits in the Rio Andres Mountains contributes to streamflow that Grande Valley are principally ET of infiltrated river infiltrates the basin alluvium shortly after flowing from water and applied irrigation water and seepage to mountain canyons onto the Tularosa Basin floor. Some agricultural drains and the Rio Grande. of this infiltration water probably recharges the regional ground-water system. Ground-water levels in Ground-Water Withdrawals the Tularosa Basin (McLean, 1970) indicate that regional flow is to the south into the Hueco Bolson in Records of historical pumpage from all known municipal supply, military, industrial, and private wells Texas and to Lake Lucero, near White Sands Missile were compiled as part of this study. For wells in Range (fig. 1). Ground-water flow from the Tularosa Mexico, annual pumpage summaries by well from Basin into the study area is probably a major 1926 through 1995 were obtained from the Junta component of recharge to the Hueco Bolson. Municipal de Agua y Sanimiento through the IBWC. Ground-water underflow from the Mesilla Basin Monthly summaries of pumpage by well for EPWU to the Hueco Bolson may occur adjacent to the Rio production wells and some military and industrial wells Grande through The Narrows. Slichter (1905) after 1967 were available. determined the alluvial thickness to be less than 26 m Historical withdrawals from the Hueco Bolson, in this area and estimated underflow to be less than 270 3 compiled from annual well pumpage records, are m /d (80 acre-ft/yr). depicted in figure 5. Beginning in the 1950’s, ground- water pumpage accelerated and reached a maximum of Human-Induced Recharge about 230 million m3/yr (186,000 acre-ft/yr) by the late 1980’s. The rapid growth of Ciudad Juarez since the Land-use zoning maps indicate that agricultural 1970’s is mirrored by increased ground-water acreage in the United States part of the Rio Grande withdrawals in Mexico. Ground-water pumping in the Valley within the model area is approximately 227 km2 United States decreased during the 1990’s as the City (56,000 acres). About 1.2 m (4 ft) of irrigation water is of El Paso began to use more treated water from the Rio applied to this acreage per year, of which 20 percent, or Grande and ground water from the Mesilla Basin. about 0.25 m (0.8 ft/yr), is estimated to infiltrate to the water table (Al Blair, oral commun., 2000). Much of Evapotranspiration this irrigation-return flow is intercepted by agricultural drains and returned to the Rio Grande. The measured pan-evaporation rate from 1950 to Seepage from the Rio Grande is a major recharge 1980 at Ysleta Yard in El Paso was 4.9 mm/d (5.8 ft/yr) component to the surrounding shallow aquifer system. (Al Blair, written commun., 2000). Although measured Much of this water probably is subsequently pan evaporation is useful for estimating potential ET discharged by evapotranspiration (ET) from the when the water table is near land surface, actual relatively shallow water table in the El Paso Valley. maximum ET rates may be higher or lower. Measured maximum ET rates for phreatophytes near the Pecos From 1948 to 1952, EPWU tested injection of River in New Mexico (Weeks and others, 1987) and the water into the Hueco Bolson aquifer system through a Gila River in Arizona (Culler and others, 1982) ranged well in El Paso Valley (Roger Sperka, El Paso Water from 1 to 4 mm/d. Phreatophyte roots may extend to 9 Utilities, oral commun., 1998). From 1971 through m deep for cottonwood and greater for mesquite 1977, several valley wells were injected at a total (Robinson, 1958). average rate of about 700 m3/d (200 acre-ft/yr). Beginning in 1981 and continuing through the 1990’s, Seepage to Rio Grande and Agricultural Drains a major artificial recharge project north of El Paso injected 10 recharge wells at rates as high as 18,000 When shallow ground-water levels are higher m3/d (5,300 acre-ft/yr). than the water level in the adjacent Rio Grande, shallow
13 ground water seeps into the Rio Grande where there is selected. For this purpose, the Groschen (1994) model sufficient hydraulic conductance. Such ground-water was used for preliminary flow-system analysis. inflow may occur southeast of El Paso, especially Running several density-dependent simulations using where shallow ground-water levels may be high from HST3D tested the magnitude of density-driven ground- applied irrigation water and insufficient drainage from water flow from heterogeneous solute concentrations. irrigated fields. The magnitude of seepage to These profile models were run with identical boundary agricultural drains can be estimated from flow measured in these drains. The BOR maintained records conditions and hydraulic properties using alternate of this flow from 1960 through 1983 (Bureau of assumptions of homogeneous or heterogeneous Reclamation, El Paso Field Division, written commun., ground-water density. For the brackish to moderately 2000). saline conditions present in the Hueco aquifer system, computed differences between the homogeneous and heterogeneous density assumptions were noticeable for Land Subsidence only a test case of isotropic hydraulic conductivity. For horizontal to vertical hydraulic-conductivity ratios Elastic aquifer compression occurs when associated aquifer pore pressure is reduced by ground- greater than 10:1 to 100:1, which are probable for water withdrawals and generally results in relatively Hueco Bolson deposits, density-driven flow effects are small but measurable land subsidence. If ground-water negligible. For the purposes of this study, the levels decline in such a way that effective stresses in assumption of constant ground-water density was aquifer-system matrix materials exceed previous concluded to be an appropriate approximation of the maximum magnitudes (the “preconsolidation stress physics of ground-water flow. level”), inelastic aquifer-system compaction may Because of the large size of this ground-water occur. For comparable incremental ground-water-level flow simulation and the need for effective model declines, inelastic aquifer-system compaction is calibration, three versions of the numerical model of typically one to two orders of magnitude greater than the Hueco Bolson were constructed: elastic compression and may result in substantial land subsidence (Riley, 1998). (1) A MODFLOW model with 1 steady-state and First-order, first-class geodetic level lines were 96 annual transient stress periods, surveyed by the National Geodetic Survey in the Hueco (2) A MODFLOWP model with 96 annual stress Bolson in 1953, 1981, and 1993. By assuming stable periods, and benchmark points on bedrock in the Hueco and (3) A MODFLOW model with 66 annual and Franklin Mountains, the relative change in benchmark 408 monthly stress periods. altitudes was determined for points in the Hueco The MODFLOW model (version 1) was first Bolson for 1953-81 and 1981-93. As of 1993, the constructed to build a reasonable simulation of the maximum measured altitude change at a benchmark near downtown El Paso was 0.25 m (0.82 ft) (Emery Hueco Bolson. An equivalent MODFLOWP model Balasz, National Geodetic Survey, written commun., (version 2) was subsequently constructed to enable 1994). This magnitude of altitude change is consistent model calibration by the inverse method (Hill, 1992). with elastic compression of the aquifer matrix resulting These two versions were frequently run with equivalent from measured ground-water drawdowns. Evidence parameter definitions and distributions to ensure model that preconsolidation levels have been exceeded has equivalency and to provide error checking. After an not been documented in the Hueco Bolson. optimal parameter set had been determined with MODFLOWP and translated to an equivalent MODFLOW input, a MODFLOW (version 3) STEADY-STATE AND TRANSIENT simulation with refined monthly discretization was run. GROUND-WATER FLOW MODEL The monthly MODFLOW simulation was checked for Because brackish to saline ground water is general equivalency to the results of the annual present in the Hueco Bolson, the significance of MODFLOW and MODFLOWP simulations. The density-driven flow effects was evaluated before the minor differences in hydraulic heads and budgets are numerical code to be used for model development was attributable to the difference in temporal discretization.
14 Numerical Method Neither MODFLOWP nor MODFLOW simulates stream leakage in model cells that are dry, Both MODFLOW and MODFLOWP use the although the actual stream channels represented in the finite-difference method to solve a form of the ground- model continue to leak at a constant rate as the water water flow equation: table declines. The stream package (Prudic, 1989) was ∂ ∂h ∂ ∂h ∂ ∂h ∂h modified in both codes to allow continuing stream K ++K K – W = S ∂x xx∂x ∂y yy∂y ∂z zz∂z ∂t s (1) leakage to the aquifer following the procedure used in the recharge package, in which areally distributed where recharge enters the topmost active cell if the upper
Kxx, Kyy, and Kzz = principal components of the layers in the model are dry (fig. 9B). This modification hydraulic-conductivity tensor; also required changes to the computation of stream h = hydraulic head; leakage reported in both the water budget and flow W = source/sink; observations computed by MODFLOWP. t = time; and Ss = specific storage. Multi-Aquifer Wells Pumping from the Hueco Bolson aquifer since The screened intervals of pumped wells in the 1903 has lowered water levels by more than 50 m (197 Hueco Bolson average more than 100 m; most of these ft) in some areas of El Paso and Juarez, dewatering part wells are screened in more than one model layer. The of the modeled area that corresponds to the top two MAW package (McDonald, 1984) was used to model layers. This situation is represented in compute discharge from each model layer (q ) to a well MODFLOW by removal of dewatered (dry) cells from k using the total discharge (Q ) specified for the well. the simulation and conversion of the underlying cells w from confined to unconfined conditions by increasing The method used in the MAW package was described the value of the storage coefficient. Simulation of by Bennett and others (1982) and has been previously dewatering in the upper part of the Hueco Bolson applied in ground-water flow models developed by aquifer required changes to MODFLOWP and Kontis and Mandle (1988) and by Groschen (1994). MODFLOW, however. Both codes were further The method is based on the Thiem (1906) equation modified by incorporating the multi-aquifer well describing steady-state radial flow to a discharging package (MAW) (McDonald, 1984), which computes well: flow to pumped wells screened in more than one model layer. These modifications are detailed in the Q r appendixes and summarized in the next two hh– = ------w ln ----- (2) w 2πT r subsections. w where h = hydraulic head at a distance r from the Aquifer Dewatering well [L]; MODFLOWP allows spatial and temporal hw = hydraulic head at the well radius rw interpolation in the computation of hydraulic head at [L]; 3 specific locations in the modeled area and also Qw = well discharge [L T]; and computes mean head in observation wells screened in T = transmissivity [L2T]. more than one model layer. However, the interpolation procedure used to compute heads in these multilayer Prickett and Lonnquist (1971) and Trescott and observation wells does not support the dewatering Larson (1976) used the Theim equation in a finite- condition described in the previous section. The difference ground-water flow model to estimate h in a interpolation procedure defined by Hill (1992) was w discharging well from the head computed at a cell modified by omitting dry layers screened by a multilayer observation well and using only the representing the well. In their applications, h was remaining saturated layers in the computation of head assumed to be equivalent to the head at an effective in the well (hw) (fig. 9A). radial distance (ra) from a well located at the center of
15 Model Model layer: layer: Water table 1 1
2 h p = 0.25 2 2 2 Water table
3 h3 p3 = 0.375 3 h3 p3 = 0.5
4 h4 p4 = 0.375 4 h4 p4 = 0.5
Before dewatering: After dewatering: hw = p2h2 + p3h3 +p4h4 EXPLANATION hw = p3h3 +p4h4 pk = fraction of the well screened in layer k hk = head in layer k hw = head in well (A) Computation of head in partially dewatered, multilayer observation wells.
Stream leakage Stream leakage hstr hstr
Model botstr Model botstr layer: layer: 1 1 Water table
Water table 2 2
Before dewatering, leakage to model layer 1: After dewatering, leakage to model layer 2: q = C(h -bot ) str str str EXPLANATION qstr = C(hstr -botstr) hstr = stream stage botstr = altitude of streambed bottom C = streambed conductance qstr = flow in stream (B) Stream leakage through dewatered model cell.
Figure 9. Modifications to MODFLOWP and MODFLOW.
16 a square cell with length (∆x) (fig. 9C). Bennett and where the subscripts = row i, column j, and layer k and others (1982) approximated the effective radius as: Qw = the algebraic sum of discharges to the well from layers m through n. Discharge to the well from each ∆x r = ------(3) model layer (qwk) can then be computed from the a 4.81 Theim equation (eq. 2) by substituting values for hw and showed that hw in a multi-aquifer well at cell ij and hijk. McDonald (1984) incorporated this method in screened in model layers m to n can be computed from: the MAW package that was written for use with n T ijkhijk MODFLOW. The MAW package was modified for this ∑ ------()⁄ - (4) study to support dewatering of model layers by ln rak rw km= Qw hw = ------n – ------n omitting dry layers from the computation of hw in T T ijk π ijk equation 4 and apportioning flows qwk to or from the ∑ ------()⁄ - 2 ∑ ------()⁄ - ln rak rw ln rak rw well in the remaining saturated layers. km= km=
Qw r a Model h1 layer:
qw1 1
h2
2 qw2 hw
h3 3 qw3
h4 4 qw4
Qw + qw3 + qw4 = qw1 + qw2 EXPLANATION
qwk = flow to or from the well in layer k r a = effective well radius Q w = well pumpage (C) Distribution of hydraulic head and flow near multilayer pumped well.
Figure 9. Modifications to MODFLOWP and MODFLOW: (figure 9C modified from Bennett and others, 1982)--Concluded.
17 The assumption of radial flow to a single well layers 1 through 9, each of which is 30 m thick. The located in the center of a model cell limits the accuracy lacustrine-playa facies is represented along the eastern of the method’s representation of certain conditions. margin of the basin in model layers 1 through 9 and The assumption of radial flow is not strictly satisfied at throughout the model in layer 10, which varies in model cells adjacent to an impermeable boundary, so thickness from 0 to 276 m. wells in these cells are not represented in the MAW package. Kuniansky and Hillestad (1980) showed, however, that if a well is located in the center of a cell Vertical Datums adjacent to an impermeable boundary, the error in hw using the ra value computed with equation 3 is less than The El Paso area has six different vertical 5 percent of the exact value computed with an datums, three of which were encountered in data used analytical solution. to define altitudes of various model boundaries. Another potential problem is representing Measured hydraulic heads were referenced to the closely spaced, multilayer wells with combined National Geodetic Vertical Datum of 1929 (NGVD- pumpage that produces an actual head (hw) lower than 29), which was also used as the vertical datum for this the value computed by the MAW package. The study. Altitudes of the Rio Grande channel surveyed by combined drawdown produced by closely spaced wells the IBWC were referenced to the IBWC datum; these could be computed by the use of smaller model cells to were converted to NGVD-29 by adding 0.348 m before more accurately represent each well or superposition streambed altitudes in the model were defined. (Reilly and others, 1984) by combining the discharge Altitudes of agricultural drains surveyed by the BOR or recharge from multiple wells into a single well were referenced to an old datum of the Santa Fe within the model cell. The head (h) computed for a Railway; these were converted to NGVD-29 by adding model cell in several test cases was determined to be 12.891 m before drain-bed altitudes in the model were relatively insensitive to the choice of these alternative defined. The differences among vertical datums used to methods, however, so multi-aquifer wells were perform these conversions are listed in table 2. specified individually in this model. Table 2. Conversions between vertical datums in the El Paso, Texas, area Spatial Discretization National International Geodetic To analyze pumping effects of individual wells, Boundary Vertical the finite-difference model grid was designed to and Water Datum of maximize spatial detail yet retain reasonable execution Santa Fe Commission 1929 time. Unconsolidated deposits ranging from 138 to 693 Datum Railway (IBWC) (NGVD-29) m thick above an altitude of 600 m are represented with Santa Fe + 12.543 + 12.891 Railway 0 meters meters 10 model layers. Each model layer consists of 165 rows - 12.543 and 100 columns of cells 500 or 1,000 m on either side. IBWC meters 0 + 0.348 meter Model layer 1 contains 10,895 active cells and -12.891 2 represents a larger area (5,099 km ) than deeper model NGVD-29 meters -0.348 meter 0 layers because the basin narrows with depth (fig. 3); layer 10 (the bottom layer) contains 8,809 active cells and represents an area of 3,839 km2. To maintain an Temporal Discretization approximately constant depth below land surface, the model-grid altitudes of all model layers increase to the A steady-state simulation was used to represent north with a gradient of 1:1,000 to the approximate predevelopment conditions and to obtain starting heads location of the Texas-New Mexico State line, north of for the transient simulation. Sixty-six annual stress which the layers are horizontal. periods and time steps were used to simulate 1903 Recent alluvial deposits in the Rio Grande Valley through 1968. Because the model parameter- are represented in model layer 1, which has an average estimation process required thousands of model runs, thickness of 30 m (fig. 3). Along the western margin of using monthly stress periods during the model- the basin, the fluvial facies is represented in model calibration process was not possible. A MODFLOWP
18 simulation with steady-state and 93 annual stress applied along arroyo channels on the east side of the periods was constructed to simulate the time period of basin and at the base of the Organ and Franklin calibration data, which extended through 1995. The Mountains in the United States and a 14-km2 (5.4-mi2) optimal parameter set obtained with MODFLOWP was area at the base of the Sierra Juarez in Mexico. The used to generate an annual stress-period MODFLOW estimate of mountain-front recharge (800 m3/d) is data set so that both models could be run and checked smaller than values used in previous modeling studies for equivalency. The 1968 head distribution from the in the area (Meyer, 1976; Groschen, 1994). During annual stress-period MODFLOW simulation was used as a starting head distribution for the monthly stress- simulation stress periods after 1924, recharge also is period simulation, which represented 1969 through applied to model cells that represent irrigated fields in 2002. the United States and Mexico. The maximum recharge Seasonal variations in Rio Grande flow, ground- rate represents the volume of water typically applied in water pumping, irrigation recharge, and ET cause excess of crop requirements in the United States (Al hydraulic-head variations that cannot be represented Blair, oral commun., 2000). During drought and low- with annual stress periods. To accurately incorporate flow years, this rate was scaled so that applied these seasonal effects, a MODFLOW simulation irrigation water would not exceed the total surface (version 3) was discretized with 408 monthly stress water available. The types and acreage of irrigated periods from January 1969 through December 2002. crops have changed during the past 100 years, but little This simulation was identical to the annual information is available to document these changes MODFLOW (version 1) and MODFLOWP over the entire simulation period, particularly in simulations (version 2) for the first 66 annual stress Mexico. The irrigated area (approximately 180 km2) periods, which simulate 1903-68. Parameter values and distributions, as well as annual total pumpage stresses, and maximum rate of return flow are, therefore, were identical to those in the annual simulations. considered constant in the model simulation. Monthly temporal discretization enabled more Specified flows represent underflow from the accurate calculations of Rio Grande seepage losses Tularosa Basin along the northern model boundary and along the 15-km (9.3-mi) reach between the end of the from the Mesilla Basin (Slichter, 1905). Ground-water lined section of the Chamizal zone and Riverside Dam. withdrawals through pumped wells are specified in model layers 1 through 9 using 1903-96 annual pumpage rates in 424 wells. Thirty-two wells that are Boundary Conditions screened within only one model layer are represented with the MODFLOW well package, and the remaining Specified-flow boundaries represent recharge 392 wells are represented with the MAW package along the basin margins and beneath irrigated fields and discharge from pumped wells screened in the fluvial described earlier. About 60 percent of the wells are facies (figs. 3 and 4). Head-dependent boundaries located in the United States, including 10 wells that are represent seepage losses from stream channels, used to recharge water to the bolson aquifer. including the Rio Grande and several irrigation canals, and ground-water discharge to stream channels, Head-Dependent Flow Boundaries agricultural drains, and ET. The exchange of water between stream channels and recent alluvial sediment in the Rio Grande Valley Specified-Flow Boundaries is simulated with the stream package in MODFLOW. Specified-flow boundaries in model layer 1 Ground-water discharge to agricultural drains is represent infiltration from ephemeral streams draining simulated with the drain package. In both packages the the mountains that border the Hueco Bolson and flow of water between the boundary and the underlying leakage from irrigation-return flows. In model layers 1 model cell is a function of the head in the stream or through 9, specified-flow boundaries represent drain, the head in the model cell, and the hydraulic underflow from upgradient areas in the Tularosa and conductance of the sediment in the channel bed: Mesilla Basins. Rates of recharge and underflow were included in the estimated parameter set considered in kA the nonlinear regression. Mountain-front recharge is C = ------(5) b
19 where k = vertical hydraulic conductivity of the bed altitudes throughout the model were obtained from sediment [LT -1]; 28-m DEM’s. Maximum depths of ET between 2.5 and A = area of the streambed or drain bed in the 9 m were tested during model calibration. Although the cell [L2]; and model was not particularly sensitive to ET extinction b = thickness of the bed sediment [L]. depth, the best model fit had an ET extinction depth of 5 m. The initial estimate (4 mm/d) of the maximum ET Geographic information system (GIS) coverages rate was refined through the regression process to a of the Rio Grande, canals, and drain channels enabled final value of 4.6 mm/d. accurate calculation of the length of these features within each model cell. Surveyed widths of the Rio Streamflow Routing Grande between the Chamizal zone and Riverside Dam (Dr. Rong Kuo, U.S. International Boundary and Water As calculated with the stream package in Commission, written commun., 1999) were used to MODFLOW, water seepage between the Rio Grande calculate the streambed area (A) of the Rio Grande and the underlying aquifer is dependent on (1) the channel for corresponding model cells. The average of difference in hydraulic head between the boundary these widths (30 m) was used to calculate streambed reach representing the Rio Grande and the model cell area for reaches where surveyed widths were representing the underlying aquifer and (2) the unavailable. The portion of agricultural drains through hydraulic conductance between the two. which water seeps (wetted perimeter) is variable but Boundary heads representing water levels of the was estimated during field reconnaissance to average Rio Grande, Franklin Canal, Ascarate wasteway, and about 3 m. Altitudes of drain-bed end points surveyed Acequia Madre are computed with the Manning by the BOR in 1960 were assigned to the associated equation in the stream package (Prudic, 1989). drain in a GIS coverage. Altitudes of intermediate Streambed slope, width, and roughness are spatially points along each individual drain were determined by variable properties specified for each boundary reach. linear interpolation along the arc length between the Stream discharge, or flow in a particular boundary end points. The drain-bed altitude assigned to each reach, varies both spatially and temporally. The slope flow-model cell was the average altitude of all drains of each stream reach was computed from surveyed within that cell. The reasonability of these altitudes was altitudes, if available, or interpolated from digital verified by comparing them with cell-average values elevation data. Manning’s n was specified as 0.03 for derived from digital elevation models (DEM’s), the Rio Grande channel and 0.004 for the Franklin corrected for an assumed drain-bed depth of 2-3 m. Canal, Ascarate wasteway, and Acequia Madre. The Because drains may have filled in with sediment since time-varying flow in the first reach of stream segments representing the Rio Grande, Franklin Canal, Ascarate 1960, drain-bed altitudes in some model runs were wasteway, and Acequia Madre was specified in the increased by 1 m during the calibration process. stream package in MODFLOW. The stream stage in Channel-bed thickness of the Rio Grande, each of these starting stream reaches was computed canals, and drains was assumed to be about 1 m. from this flow using the Manning equation. Discharge Because the bed thickness is somewhat uncertain and into each subsequent downstream stream reach was wetted perimeter widths vary, the vertical conductance adjusted by seepage loss or gain from the connected per unit length of each of these features, rather than the aquifer model cell, and stream stage in each reach was vertical hydraulic conductivity, was included in the calculated with the Manning equation. estimated parameter set considered in the nonlinear Surface-water flows in the Rio Grande, regression. American Canal, Franklin Canal, Ascarate wasteway, In the ground-water flow simulations, essentially and Acequia Madre in Mexico were simulated with nonirrigated and undrained agricultural conditions seven stream segments in the stream package (fig. 8). change to irrigated and drained conditions in the Rio Diversions from the Rio Grande flow into the Franklin Grande Valley in 1925. (For ground-water seepage to Canal and Acequia Madre. A diversion from the the 21 major agricultural drains simulated with the Franklin Canal back to the Rio Grande flows in the drain package in MODFLOW and MODFLOWP, the Ascarate wasteway. Flow in the Franklin Canal is reader is referred to figure 18D.) routed back to the Rio Grande through the Tornillo ET from nonirrigated riparian land in the Rio Canal. Grande Valley was represented using the ET package Daily records of flow in the Rio Grande at of MODFLOW and MODFLOWP. Land-surface American Dam, diversion into the Franklin Canal, and
20 diversion to Mexico at International Dam from June 1, Qleon = flow in the Leon wasteway; 1938, through December 31, 1996, obtained from the Qhask = return flow from the Haskell Waste IBWC (2000) were used to compute mean annual and Water Treatment Plant; and mean monthly input flow specifications for stream Qevap = estimated total evaporation from segments representing the Rio Grande, American the American Canal between Canal, Franklin Canal, and Acequia Madre. Water American Dam and the end of the diverted from the Franklin Canal back to the Rio Chamizal zone. Grande was specified according to measurements Because tributary inflows from the Settling obtained from the BOR (El Paso Field Division, Basin wasteway and Leon wasteway were not directly written commun., 2000). measured, their inflow was derived by subtracting the Mean annual flows in the Acequia Madre, measured flow in Franklin Canal and the diversion to Franklin Canal, and Rio Grande from 1903 through the Canal Water Treatment Plant from the measured 1996 are shown in figure 10A. Diversions from the Rio flow in Franklin Canal downstream from American Grande into the Franklin Canal began in 1938. Flow in Dam. In equation form: the Rio Grande was virtually nonexistent during the drought years of 1954-57. The late 1970’s were Q + Q = Q – Q –Q (7) relatively dry years; flow during this period is shown in sb leon frank1 wtp frank2 monthly detail in figure 10B. In contrast, the late where Q = measured flow in the Franklin 1980’s to early 1990’s were relatively wet; flow during frank1 this period is shown in monthly detail in figure 10C. Canal below American Dam; Flow in the stream segment simulating the Rio Qwtp = measured diversion to the Canal Grande at the end of the Chamizal zone was specified Water Treatment Plant; and for the monthly stress periods beginning in January Qfrank2 = measured flow in the Franklin 1969. Because Rio Grande flow was measured Canal between the Leon and upstream from the lined section of the American Canal, Ascarate wasteways. the flow specified for the simulation required Monthly return flows from the Haskell Waste Water adjustment to account for tributary inflow and loss Treatment Plant (Qhask) were relatively constant between the streamflow-gaging station and the start of (Roger Sperka, oral commun., June 2000), enabling 3 the unlined section of the Rio Grande. These use of a constant average flow of 75,708 m /d. The adjustments were made according to an analysis of the monthly evaporative fluxes (Qevap) were generated El Paso County Water Improvement District canal from local pan-evaporation rates by Al Blair (written system (Al Blair, written commun., January 26, 2000). commun., 2000). Additions to the flow measured downstream from To specify monthly flow in the Rio Grande from American Dam were made to account for inflow from 1994 through 2004, a “design flow” was synthesized the Settling Basin wasteway, Leon wasteway, and from other components of the El Paso surface-water Haskell Waste Water Treatment Plant. The budget (Al Blair, written commun., 2000). The accumulated flow was debited to account for the specified flow for the stream-package segment diversion to Mexico at International Dam and representing the Rio Grande downstream from the estimated evaporation in the canal between American Chamizal zone was set equal to 60 percent of the design Dam and the downstream end of the Chamizal zone. diversion to El Paso plus the design diversion at This flow is summarized in equation form: Riverside plus the computed seepage loss in the 15-km reach between the Chamizal zone and Riverside Dam. (Because the seepage loss in this segment was = – ++ + – Qseg Qamer Qmex Qsb Qleon Qhask Qevap (6) computed with the ground-water flow model, which required this flow specification as input, several where Qseg = specified inflow to the stream iterations of the model were run to obtain a flow segment below the Chamizal specification.) This calculation may be summarized as: zone; Q = measured flow in the American amer () Canal below American Dam; RioGrande = 0.6 diversion ++seepage Riverside˙ (8) Qmex = measured diversion to Mexico at International Dam; The resulting design-flow specifications for stream Qsb = flow in the Settling Basin segments representing the Acequia Madre, Franklin wasteway; Canal, and Rio Grande are shown in figure 10D.
21 9,000 (A) 1903-96, by year 8,000 7,000 6,000 5,000 4,000 3,000
2,000 1,000 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990
5,000 (B) 1975-79, by month 4,000
3,000
2,000
1,000
0 1975 1976 1977 1978 1979
5,000 (C) 1988-92, by month 4,000
3,000
, i n t h o u s a n d s o f c u b i c m e t e r s p e r d a y , i n t h o u s a d f c b m e r p 2,000
1,000
0 1988 1989 1990 1991 1992 S t r e a m f l o w
5,000 (D) Design flow, by month 4,000 3,000
2,000 1,000
0 Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
EXPLANATION Acequia Madre Franklin Canal Rio Grande
Figure 10. Mean annual and mean monthly flows in Rio Grande and diversions, 1903-96.
22 Faults flow losses or gains compose the calibration data set (fig. 12). Fault gouge in subvertically dipping fault planes may impede horizontal fluid flow and affect ground- water flow patterns (Haneberg and others, 1999). Hydraulic-Head Measurements Several intrabasin faults deforming Quaternary alluvium are recognized northeast of El Paso (Barnes, The nonlinear regression included 4,352 heads 1983) (fig. 11). To evaluate the influence of these faults measured between 1912 and 1995 in 310 wells. These on ground-water flow, they were incorporated into the measurements were compiled from databases of ground-water flow simulation with the Horizontal Flow EPWU and the U.S. Geological Survey (USGS). About Barrier package (Hsieh, 1992). 83 percent (3,615) of the measurements were made in 226 wells in the United States, and the remainder (737) were made in 84 wells in Mexico. Most of the heads Storage Properties have been measured since ground-water withdrawals increased significantly in the 1960’s (fig. 12). Of the Water in storage in the aquifer system may be total set of measurements, 4,184 were made in wells released by three main processes: (1) drainage of pore with screened intervals corresponding to more than one space in unconfined layers, (2) expansion of water, and model layer; these were defined as multilayer head (3) compression of the aquifer matrix. Drainage of measurements in MODFLOWP. water from pore space above a declining water table is the dominant process yielding water to the ground- water flow system and is quantified by specific yield Flow Measurements (Sy). Specific yield generally ranges from 10 to 20 percent. Its magnitude in the Hueco Bolson was Measurements of streamflow loss in the Rio estimated in the model-calibration process as a uniform Grande and Franklin Canal and seepage into the Island value throughout the alluvial deposits. Specific yield Drain were used to constrain the nonlinear regression. and the component of specific storage that is due to the The IBWC computed seepage loss from the Rio expansion of water (which is very minor) were Grande into the underlying aquifer along a 15-km (9.3- simulated with the Block-Centered-Flow package in mi) reach between the concrete-lined section in the MODFLOW and MODFLOWP. The skeletal Chamizal zone and Riverside Dam for 1981, 1982, and components of specific storage, which quantify water 1983 (Land and Armstrong, 1985; White and others, derived from compression of the aquifer-system 1997). The BOR computed flow loss along an 8.5-km matrix, were simulated with the Interbed-Storage (5.25-mi) reach of the Franklin Canal for 1984, and the package in MODFLOW and MODFLOWP. The USGS computed flow loss for 1990 through 1992 magnitude of elastic skeletal specific storage (S ) is ske (Land and Armstrong, 1985; White and others, 1997). reasonably well constrained by plausible physical These computations and simulated flow losses are properties of alluvial deposits. In the El Paso area, summarized in table 1. onsite estimates of S (Heywood, 1995) have been ske Although flow measured in some agricultural made from piezometric-extensometric measurements; -6 -1 drains may include tributary inflow, such inflow to the this value (Sske =7x10 m ) was specified in the flow model. Island Drain is minor. Twenty-four measurements of flow in the Island Drain were used to constrain drain- bed hydraulic conductance in the regression. MODEL CALIBRATION
MODFLOWP computes the changing spatial Parameter Estimation distribution of ground-water heads and fluxes across head-dependent boundaries, which were used to Parameter values representing aquifer properties represent the Rio Grande, irrigation canals, and and specified boundaries were estimated through agricultural drains. During model calibration, nonlinear regression that minimized differences parameter values were adjusted to minimize between measured and computed heads and flows in a differences between computed and measured heads and 94-year transient-state simulation from 1903 through fluxes. A total of 4,439 measurements of heads and 1996. Hydraulic heads computed in a steady-state
23 106˚30´ 106˚22´30´´ 106˚15´
32˚
405 14 15
30
31˚ 52´ 30´´ 413
416 43
59
86 85
104
124 F ran 123 141 klin 31˚ 148 45´ 176 C a n a l
202 Rio 209 207 212 G ra Acequia nde
Madre
Shaded-relief map from U.S. Geological Survey 0 2 4 6 8 10 KILOMETERS 30-meter Digital Elevation Models 0 24 6 8 10 MILES EXPLANATION Active model-grid boundary Location of wells in figures 6 and 7
Agricultural drains Location of wells with hydrographs in figure 17
Faults as represented in flow model Location of other water-level measurement sites
Figure 11. Locations of head measurements, stream segments, agricultural drains, and model faults.
24 160
United States 140 Mexico
120
100
80
60 Number of measurements
40
20
0 1903 1908 1913 1918 1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 Figure 12. Number of annual head measurementsin wells included in the nonlinear regression used in the ground-water flow model. simulation representing prepumping conditions, when Improvements in model results were identified no large stresses were imposed on the aquifer system, by comparing the sum of squared errors (SSE): provide initial conditions for the transient-state simulation. The nonlinear regression procedure was Σ[]12⁄ 2 SSE = wi ei , i = 1, n (9) implemented using UCODE, in which sensitivities of model parameters are estimated through a where ei = difference between measured and perturbation technique (Poeter and Hill, 1998). calculated values of Although MODFLOWP provides a similar regression measurement i; 1/2 procedure in which sensitivities of model parameters wi = square root of the weight are computed directly from an analytical expression, assigned to the error in the a major limitation prevented its application for model measured value of calibration. Conversion from confined to unconfined measurement i; 1/2 conditions caused by dewatering the upper part of the wi ei = weighted residual corresponding modeled area can be represented in MODFLOWP in to measurement i; and the solution for hydraulic head, but this conversion was n = number of measurements; not implemented in MODFLOWP for the regression and the standard error of procedure. The calibration procedure selected to estimate (SEE): circumvent this problem entailed (1) using SSE 12⁄ MODFLOWP to define the parameter values in the SEE = ------, (10) model and compute hydraulic heads and flows and (2) np– running UCODE to estimate the optimum parameter where p = number of model parameters estimated by values. the regression.
25 The weights, wi, were chosen according to slightly. The “block” model was found to have lower procedures in Hill (1992) to account for the different overall error; the geometry of this hydraulic- units associated with head measurements (m) and flow conductivity distribution is depicted in cut-away measurements (m3/d). The water-level altitudes perspective in figure 4. Although the horizontal reported for wells in the United States used a datum hydraulic conductivities of the alluvial-fan and fluvial consistent with other data in the flow model. Heads in facies were independently estimated, they converged to the United States were generally measured several days essentially identical values. This suggests that the after the cessation of pumping in wells to better location of this facies boundary may not be significant represent nonpumping conditions. The time elapsed in the context of the ground-water flow model. after cessation of pumping is not known for wells in The total SSE computed using equation 7 was Mexico, however, and the reported water-level altitudes 38,814 m2. The SSE for all model heads computed with are less certain. The wi values were therefore adjusted equation 8 was 2.99 m2. The SSE for heads in United so that heads in wells in Mexico were weighted 33 States wells was 2.90 m2 and for heads in Mexican percent less than heads in wells in the United States. wells was 3.44 m2. Scatter plots of simulated heads in Heads measured in the United States were weighted relation to measured heads and of simulated heads in equally. Flow measurements were weighted to reflect relation to weighted residuals for wells in Mexico and the relative accuracy of Rio Grande and canal seepage- the United States are shown in figure 13C-D. The loss measurements and drain-flow measurements. scatter plots for the United States wells show no Measurements of Rio Grande seepage loss (table 1) obvious evidence of model bias. The scatter plot of were weighted equally to heads measured in wells in the United States. Measurements of Franklin Canal simulated heads in relation to weighted residuals for seepage loss and flow in the Island Drain were wells in Mexico suggests some bias toward negative weighted 67 percent less than measurements of Rio weighted residuals. This bias indicates that heads in Grande seepage loss. Mexican wells are, in general, slightly lower than those predicted by the ground-water flow model.
MODEL EVALUATION AND SIMULATION Estimates of Aquifer Properties RESULTS The spatial distribution of horizontal and vertical The best-fit parameter values along with their hydraulic conductivities was defined by assigning 95-percent confidence intervals computed from the zones to each of the hydrogeologic facies, to which nonlinear least-squares regression of the “block” facies corresponding parameters representing hydraulic model are summarized in table 3. Because the lengths conductivities were applied. The horizontal extent of of the Rio Grande, Franklin Canal, Acequia Madre, and the recent alluvial facies (fig. 3) is known because agricultural drains in each model cell are well known deposition was constrained by the topography of the but the thickness of bed material is uncertain, these Rio Grande Valley. Boundaries between the fluvial, parameters are reported as a conductance per unit lacustrine-playa, and alluvial-fan facies are less length. Quaternary fault-zone hydraulic conductivity constrained. Two possible geometries for the boundary was estimated on the basis of a fault-zone thickness of between the fluvial and lacustrine-playa facies were 1 m. If actual fault-zone thickness is on the order of 10 tested: (1) a “block” model with a vertical facies or 1 cm, the actual fault-zone hydraulic conductivity is boundary and (2) a “wedge” model in which the facies one or two orders of magnitude smaller, respectively. boundary sloped down to the west so that lacustrine- playa facies underlie the fluvial facies in parts of layers 2 through 9. In both models, the lacustrine-playa facies Parameter Sensitivities accommodated all of layer 10 and the hydraulic conductivities between the two facies were gradational, Composite scaled sensitivities (Hill, 1998) are simulating some facies interfingering. For each model, listed in table 2 for model parameters estimated in the the parameter-estimation regression was run to nonlinear regression. These sensitivities vary determine the lowest SSE parameter set. The resulting somewhat as a function of parameter values and parameter sets for these two models differed only distribution. The sensitivities are useful for interpreting
26 Mexico wells United States wells
1,140 1,140 A B 1,130 1,130
1,120 1,120 vel vel
1,110 1,110
1,100 1,100 above sea le above above sea le above
1,090 1,090 Measured head, in meters Measured head, in meters
1,080 1,080
1,070 1,070 1,070 1,080 1,090 1,100 1,110 1,120 1,130 1,140 1,070 1,080 1,090 1,100 1,110 1,120 1,130 1,140
Simulated head, in meters above sea level Simulated head, in meters above sea level
Mexico wells United States wells 20 20 C D 15 15
10 10 us simulated) us simulated) 5 5
0 0
-5 -5
-10 -10
-15 -15
Residual head (measured min -20 Residual head (measured min -20 1,070 1,080 1,090 1,100 1,110 1,120 1,130 1,140 1,070 1,080 1,090 1,100 1,110 1,120 1,130 1,140 Simulated head, in meters above sea level Simulated head, in meters above sea level
Figure 13. Simulated and measured heads and weighted residuals computed in transient-state simulation of the Hueco Bolson.
27 Table 3. Optimum parameter values estimated for Hueco Bolson aquifer, through nonlinear regression in transient-state simulation, and their approximate 95-percent confidence intervals
[m, meters; m/d, meters per day; m3/d, cubic meters per day; N/A, not applicable]
Scaled sensitivity, Parameter Value Confidence interval in meters Recharge: 1 Irrigation-return flow 3.8 x 104 m3/d 2.2 x 104 - 6.5 x 104 m3/d 0.3 Underflow from Tularosa Basin 2.0 x 104 m3/d 1.9 x 104 - 2.2 x 104 m3/d 7 2 Mountain front on alluvial fans 8.0 x 102 m3/d 0 - 2 x 103 m3/d 2 Underflow from Mesilla Basin 3.4 x 102 m3/d Not estimated N/A Horizontal hydraulic conductivity:
Alluvial-fan facies 6.8 m/d 6.0 - 7.7 m/d 5 Recent fluvial sediments 4.0 m/d 2.8 - 7.2 m/d 2 Fluvial and alluvial facies 6.8 m/d 6.4 - 7.2 m/d 10 Lacustrine-playa facies 0.9 m/d 0.5 - 1.4 m/d 0.3 Quaternary faults 3.8 x 10-3 m/d 1 x 10-3 - 1 x 10-2 m/d 8 Vertical hydraulic conductivity:
Alluvial-fan facies 1.2 x 10-3 m/d 6 x 10-4 - 2 x 10-3 m/d 3 Recent fluvial sediments 1.3 x 10-1 m/d 6 x 10-2 - 6 x 10-1 m/d 2 Fluvial and alluvial facies 1.3 x 10-2 m/d 1 x 10-2 - 1.5 x 10-2 m/d 0.6 Lacustrine-playa facies 2.5 x 10-2 m/d 6 x 10-3 - 1 x 10-1 m/d 0.4 Specific yield 0.178 0.173 - 0.184 10 Specific storage - elastic 7 x 10-6 m-1 2 x 10-6 - 1 x 10-5 m-1 0.5 Specific storage - inelastic 7 x 10-5 m-1 Not estimated N/A Conductance per unit length:
Rio Grande 1.8 m/d 1.6 - 2.0 m/d 1 Agricultural drains 5.8 m/d 2 - 1.6 x 101 m/d 2 Irrigation canals 3.0 m/d Not estimated N/A Evapotranspiration extinction depth 5 m Not estimated N/A Maximum evapotranspiration rate 4.6 x 10-3 m/d 1 x 10-3 - 7 x 10-3 m/d 2 Manning’s n:
Rio Grande 0.03 Not estimated N/A Franklin Canal, Acequia Madre 0.004 Not estimated N/A
1Maximum for normal irrigation year after 1933. 2Distributed in arroyos below Organ, Franklin, and Juarez Mountains.
28 relative sensitivities between model parameters and the deactivation by ground-water drawdown beneath the relative importance of a parameter in reducing model bottom of model layer 1. Model layer 2 has similarly error in the nonlinear regression. Specific yield of deactivated in some areas. unconfined layers and horizontal hydraulic Hydrographs of measured and model-simulated conductivity of the fluvial facies were the most water levels for 18 wells are presented in figures sensitive model parameters, followed by recharge 17A-G. The locations of these wells, which are underflow from the Tularosa Basin and conductance of representative of the other 274 wells in the Quaternary fault zones. measurement set, are shown in figure 11. These hydrographs illustrate the general quality of model fit and may be useful for showing areas where the model Water-Level Drawdowns could be improved. (A spreadsheet containing hydrograph measurements for the remaining 274 wells Simulated water-table altitudes for 1902 (steady- is available on request.) state conditions) and for 1958, 1973, and 1980 (transient-state conditions) are contoured at 2-m intervals and shown in figure 14A-D. The progressive Ground-Water Budget growth of cones of depression under El Paso and Components of the ground-water budget that Ciudad Juarez is evident from 1958 through 1980. change during the course of the transient simulation are Shallow ground water flows away from the Rio Grande shown in figure 18. Water pumped from wells is toward these cones of depression on either side of the supplied primarily by releases from ground-water international border. storage, principally by drainage of aquifer pore space; Potentiometric surface altitudes computed for this process results in large declines of the water table. model layer 5, which is approximately 135 m (440 ft) The mirror relation between ground-water pumping below the steady-state water table, for 1958, 1973, and releases from storage is evident in figure 18A,B. A 1980, and 1996 are shown in figure 15A-D. This model similar mirrorlike relation can be seen in figure 18C,D, layer is in the middle of the depth interval in which which illustrates the principal budget components in many production wells are screened in the Hueco the shallow portion of the aquifer system in the Rio Bolson. The contours for 1958 through 1996 show the Grande Valley. These two graphs indicate that Rio expanding cones of depression under El Paso and Grande water infiltrating into the shallow aquifer Ciudad Juarez; the Rio Grande, however, has much less system is consumed primarily by ET and (or) flows to influence at this depth than at the water table. Since agricultural drains. about 1980 (fig. 15C), water at this depth generally flows from north to south beneath the river toward the lower potentiometric heads beneath Ciudad Juarez. By Seepage from the Rio Grande 1973, ground-water drawdown under Ciudad Juarez had created a ground-water divide to the southeast, The simulated seepage losses from the Rio which isolated ground-water flow near the city from the Grande between the end of the Chamizal zone and regional pattern of southeast flow. This divide has Riverside Dam for two historical and one hypothetical migrated farther to the southeast as the cone of (design flow) 5-year periods are depicted in figure depression has deepened. The influence of simulated 19A-D. Although model fluxes and other report figures Quaternary faults (fig. 11) becomes increasingly have units in meters per day, these seepage losses are evident as deflections of water-level contours in later presented in acre-ft/yr for the benefit of interested transient stress periods, such as depicted for 1980 and parties. Variations in seepage by month from 1975 to 1996 (fig. 15C,D). 1979 and 1988 to 1992 are shown in figure 19A. The three-dimensional pattern of ground-water Annual seepage loss was divided into seepage during drawdown beneath El Paso and Ciudad Juarez is the primary and secondary irrigation seasons in figure further illustrated by perspective views for 1973 and 19B-D. Seepage losses for 1975-79 and 1988-92 were 1996 (fig. 16A,B). The vertical section of these images during times of relatively low and high flow in the Rio was sliced along an azimuth of N. 30° E. to transect the Grande, respectively, and correspond to the monthly drawdown cones beneath both cities. Only active periods illustrated in figure 10. The greater seepage in model cells are shown; missing cells in the El Paso and 1988-92 principally results from higher average stage Ciudad Juarez regions in 1996 (fig. 16B) result from of the Rio Grande.
29 106˚30´ 106˚15´ 106˚00´
1156 1154 32˚ 15´ 1152
1150 1148
1146
1144
1142 1140 1138 32˚ 1136 00´ 1134
1132 EXPLANATION Water-table contour--Shows 1128 altitude of water table. 1126 Contour interval 2 meters. 1124 Datum is sea level 1122 1120 1118 El Paso 1116 1114 31˚ 1112 45´ 1110 1108 1106 1104 1102 Ciudad Juarez 00 11 1098 1096 94 10 1092 1090 1088
31˚ 30´ 1086
1084
1082
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 14A. Simulated water table in shallow aquifer (model layers 1 and 2) in 1902 (steady-state conditions).
30 106˚30´ 106˚15´ 106˚00´
1156 1154 32˚ 15´ 1152 1150 1148
1146
1144
1142 1138 1136 32˚ 00´ 1134
1132 EXPLANATION 1130 Water-table contour--Shows 1128 altitude of water table. 1126 124 Contour interval 2 meters. 1 Datum is sea level 1122 1120 1118 El Paso 1116 1114 31˚ 1112 8 45´ 1120 1116 1110 0 11 06 11 1104 1118 1102 1100 Ciudad Juarez 1098 1096 1094
1092 90 10 1088
6 31˚ 108 30´
1084
1082
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 14B. Simulated water table in shallow aquifer (model layers 1 and 2) in 1958.
31 106˚30´ 106˚15´ 106˚00´
1158 1156 1154 32˚ 15´ 1152
1150 1148
1144
1140 1138 1136 32˚ 00´ 1134
1132 EXPLANATION 1130 Water-table contour--Shows 1128 altitude of water table. Contour interval 2 meters. 1126 4 Datum is sea level 1112 112 1108 1110 1122 1120 1106 1118 El Paso 1116 1114 2 31˚ 111 45´ 1110 1108 1106 1104 1102 1114 Ciudad Juarez 0 110 1116 1098 1096 1094 1092 1090 1114 1088
31˚ 30´
1086
1084 1082
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 14C. Simulated water table in shallow aquifer (model layers 1 and 2) in 1973.
32 106˚30´ 106˚15´ 106˚00´
1156 1154 32˚ 15´ 1152
1150 1148
1146
1144
114 1140 2 1138 1136 32˚ 00´ 1134
1132 1130 EXPLANATION 11 Water-table contour--Shows 28 altitude of water table. 1126 Contour interval 2 meters. 1124 Datum is sea level 1122 18 1120 11 El Paso 1116 1102 1104 1114 1106 31˚ 1108 10 1112 11 12 45´ 11 1110 1108 1106 1098 1104 1100 1116 1102 Ciudad Juarez 1100 14 11 1098 6 109 1094 1092 90 10 1088
31˚ 6 30´ 108
1084
1082
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 14D. Simulated water table in shallow aquifer (model layers 1 and 2) in 1980.
33 106˚30´ 106˚15´ 106˚00´
1154 32˚ 15´ 1152
1150
1148
1146 1144 1142
1138
1136 32˚ 00´ 1134
1132 1130 EXPLANATION 1128 Potentiometric contour–- 1126 Shows altitude at which 1124 water level would have stood in tightly cased 1122 wells. Contour interval El Paso 1120 2 meters. Datum is sea level 1118 1116 1114 1114 31˚ 1112 45´ 1110 1108 1106 1104 Ciudad Juarez 1102 00 1116 11 98 10 1096 1094 1092 1090 1088
31˚ 6 30´ 108
1084
1082
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 15A. Simulated potentiometric surface in model layer 5 in 1958.
34 106˚30´ 106˚15´ 106˚00´
11 32˚ 54 15´ 1152
1150 A' 1148
1146 1144 1142
1140
1138
1136
32˚ 1134 00´
1132
113 0 EXPLANATION 1128 Potentiometric contour–- 1126 B' Shows altitude at which 1124 water level would have 1122 stood in tightly cased wells. Contour interval 1120 El Paso 2 meters. Datum is sea level 1118 1104 1116 8 1106 0 1114 31˚ 11 1112 1110 12 45´ 11 1110 1108 1106 1104 Ciudad Juarez 1102 A 1100 1098
1096
1094 1092 8 1090 108
31˚ 30´ 1086
1084
1082
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 15B. Simulated potentiometric surface in model layer 5 in 1973.
35 106˚30´ 106˚15´ 106˚00´
1154 32˚ 15´ 1152
1150
1148
1 146 1144 1142
1140
1138 1136 32˚ 00´ 1134
1132
1130 1128 EXPLANATION 1126 Potentiometric contour–- Shows altitude at which 1124 water level would have 1122 stood in tightly cased wells. Contour interval 1120 2 meters. Datum is sea level El Paso 8 1118
09 1098 1 1116 4 1100 111 31˚ 11021104 1112 1106 110 45´ 1108 1 10 11 1108 1106 1104 02 Ciudad Juarez 11
2 1100 11 1 1098
1096 1094 92 10 1090 1088
31˚ 30´ 6 108
1084
1082
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 15C. Simulated potentiometric surface in model layer 5 in 1980.
36 106˚30´ 106˚15´ 106˚00´
32˚ 15´ 1152
1150
1148 1146 1144 1142 1140
1138 1136
32˚ 1134 00´
1132 1130
1128 1126 EXPLANATION 1124 Potentiometric contour–- Shows altitude at which 11 water level would have 22 1120 stood in tightly cased wells. Contour interval 1118 2 meters. Datum is sea level El Paso El1086 Paso 1084 1088 1116 1090 4 1078 1080 1082 111 1092 1094 31˚ 1076 1112 1096 1098 00 1102 1110 45´ 11 1104 6 1108 110 1106 1104 Juarez Ciudad Juarez 1102 1100 1098 1096
1094 1092 1090 1088
31˚ 30´ 1086
1084
Shaded-relief map from U.S. Geological Survey 30-meter Digital Elevation Models 0 5 10 15 20 25 30 35 40 KILOMETERS
0 5 10 15 20 25 30 35 40 MILES Figure 15D. Simulated potentiometric surface in model layer 5 in 1996.
37 EL PASO Logarithmic color scale. Perspective view showing water-level drawdown in the Hueco Bolson aquifer in 1973. in the Hueco Bolson aquifer drawdown water-level showing view Perspective Figure 16A. CIUDAD JUAREZ CIUDAD 60 45 30 16 1.0 Water-level drawdown, in meters
38 EL PASO arithmic color scale. g Lo Perspective view showing water-level drawdown in the Hueco Bolson aquifer in 1996. in the Hueco Bolson aquifer drawdown water-level showing view Perspective Figure 16B. CIUDAD JUAREZ CIUDAD 16 60 1.0 30 45 Water-level drawdown, in meters
39 1,135
JL-49-05-602 (well 14)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l
JL-49-06-401 (well 15)
1,125
1,115
W a t e r - l v i u d 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990
Figure 17A. Measured and simulated water levels from 1930 through 1995 in selected wells in the Mesa-Nevins well field (location of wells shown in figure 11).
40 1,135
JL-49-05-901 (well 30)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l
JL-49-13-204 (well 43)
1,125
1,115
W a t e r - l v i u d 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990 Figure 17A. Measured and simulated water levels from 1930 through 1995 in selected wells in the Mesa-Nevins well field (location of wells shown in figure 11)--Concluded.
41 1,135
JL-49-13-601 (well 59)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l JL-49-13-610 (well 86)
1,125
1,115
W a t e r - l v i u d 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990
Figure 17B. Measured and simulated water levels from 1930 through 1995 in selected wells in the Airport well field (location of wells shown in figure 11).
42 1,135
JL-49-14-413 (well 85)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l JL-49-13-804 (well 104)
1,125
1,115
1,105 W a t e r - l v i u d
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990 Figure 17C. Measured and simulated water levels from 1930 through 1995 in selected wells in the Cielo Vista and Town well fields (location of wells shown in figure 11).
43 1,135
JL-49-13-702 (well 124)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l
JL-49-13-710 (well 141)
1,125
1,115
W a t e r - l v i u d 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990
Figure 17D. Measured and simulated water levels from 1930 through 1995 in selected wells in the water-plant well field (location of wells shown in figure 11).
44 1,135
JMAS 10-R (well 176)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l
JMAS 17 (well 202)
1,125
1,115
W a t e r - l v i u d 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990 Figure 17E. Measured and simulated water levels from 1930 through 1995 in selected wells in the Juarez well field (location of wells shown in figure 11).
45 1,135
JMAS13-R (well 209)
1,125
1,115
1,105
l
1,095
l e v e
1,085
Measured Simulated
a b o v e s e a 1,075
1,135
, i n m e t e r s
e JMAS42 (well 212)
u d
1,125
a l t i t
1,115
- l e v e l
W a t e r 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990 Figure 17E. Measured and simulated water levels from 1930 through 1995 in selected wells in the Juarez well field (location of wells shown in figure 11)--Concluded.
46 1,135
JL-49-14-102 (well 413)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l
JL-49-14-301 (well 416)
1,125
1,115
W a t e r - l v i u d 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990 Figure 17F. Measured and simulated water levels from 1930 through 1995 in selected wells in the east observation well field (location of wells shown in figure 11).
47 1,135
JL-49-21-301 (well 148)
1,125
1,115
1,105
1,095
1,085
Measured Simulated
1,075
1,135 , i n m e t r s a b o v l
1,125
JL-49-05-502 (well 405)
1,115
W a t e r - l v i u d 1,105
1,095
1,085
Measured Simulated
1,075 1930 1940 1950 1960 1970 1980 1990 Figure 17G. Measured and simulated water levels from 1930 through 1995 in selected wells in the lower valley and north observation well fields (location of wells shown in figure 11).
48 0 -100 -200 -300 -400 -500 -600 (A) Ground-water withdrawals -700 -800
600 500 400 (B) Storage 300 200 100 0 -100 -200
800 (C) Stream leakage 700 600 500 400 300 200 100 0 T h o u s a n d s o f c u b i c m e t e r s p e r d a y T h o u s a n d f c b i m e t r p 0 -100 -200 -300 -400 -500 -600 (D) Agricultural drains -700 Evapotranspiration -800 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995
Figure 18. Simulated annual inflows and outflows (-) to Hueco Bolson aquifer.
49 50
40
30
20
10 (A) Monthly seepage 1975-79 1988-92 0 Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec 45 (B) 1975-79 40 35 30 25 20 15 10 5 0 Jan 1975 Jan 1976 Jan 1977 Jan 1978 Jan 1979 45 (C) 1988-92 40 35 30 25 20 15 10 5 0 Jan 1988 Jan 1989 Jan 1990 Jan 1991 Jan 1992 T h o u s a n d f c r e - t p y 45 40 35 30 25 20 15 10 (D) Rio Grande seepage with design flow 5 0 Jan Jan Jan Jan Jan year 1 year 2 year 3 year 4 year 5 EXPLANATION Total annual Primary irrigation season Secondary irrigation season
Figure 19. Rio Grande seepage between Chamizal zone and Riverside Dam (locations shown in figure 8).
50 The simulated seepage for the 10th year of a SUMMARY AND CONCLUSIONS “design flow” in the Rio Grande without the ACE is illustrated in figure 10D. This design-flow seepage-loss The neighboring cities of El Paso, Texas, and Ciudad Juarez, Chihuahua, Mexico, have historically calculation is for a hypothetical scenario in which Rio relied on ground-water withdrawals from the Hueco Grande water is not diverted into the ACE. The Bolson, an alluvial-aquifer system, to supply water to simulated aquifer conditions for this scenario their growing populations. In the United States, approximate those that would have occurred during diversions from the Rio Grande and ground-water 2002 if the monthly ground-water pumpage rates and withdrawals from the Mesilla Basin west of the study patterns in 1992 repeated for 10 years. The computed area also supply the freshwater demands of the seepage loss from the Rio Grande to the underlying military, industries, and public in the El Paso area. In aquifer along the 15-km reach between the end of the Mexico, diversions from the Rio Grande are used for Chamizal zone and Riverside Dam is 39,000 agriculture; water needed by Ciudad Juarez is supplied acre-ft/yr in this scenario. solely by extraction from the Hueco Bolson. By 1996, The altitudes of drain beds specified in the model ground-water drawdown exceeded 60 m in some areas under Ciudad Juarez and El Paso. Fresh ground water influence simulated water-table altitudes in portions of stored in the aquifer system beneath these cities is the Rio Grande Valley, which could affect seepage rates bordered by regions of brackish to saline ground water. from the Rio Grande to the underlying aquifer. As water levels in the freshwater portions of the aquifer Although uncertainty exists for the drain-bed altitudes declined, intrusion of the surrounding brackish water specified in the simulation (which are probably +1, -2 degraded water quality in public supply wells, which m), these altitudes were not formally estimated during sometimes required well abandonment. the model-calibration process. To test the possible Rio By the end of 1996, ground-water levels had Grande seepage-loss effect of the specified drain-bed declined by as much as 60 m (197 ft) from simulated altitudes, several forward model runs were made with steady-state levels in the Hueco Bolson. By drain-bed altitudes decreased in increments of 1 m. incorporating extensive pumpage records with fine Although drain seepage increased, seepage loss from spatial and temporal discretization, these declines have the Rio Grande between the end of the Chamizal zone been simulated in a numerical ground-water flow and Riverside Dam remained at 39,000 acre-ft/yr. model that matches 4,352 head observations with an SEE of about 3 m. This monthly temporal Within the tested drain-bed altitude range, which discretization was needed to (1) permit accurate brackets values deemed reasonable, seepage loss from calculations of seepage losses from the Rio Grande and this reach of the Rio Grande is insensitive to specified (2) provide a useful ground-water management model. drain-bed altitude. The simulation of steady-state and transient The calibration on both the computed Rio ground-water flow in the Hueco Bolson was developed Grande seepage losses and the adjustments to using MODFLOW-96. The transient simulation measured flow required for stream-package flow represents a period of 100 years beginning in 1903 and specification (see “Streamflow routing” section) was ending in 2002. The period 1903 through 1968 was further checked by comparing evaporation-corrected, represented with 66 annual stress periods, and the model-simulated flow at Riverside Dam with measured period 1969 through 2002 was represented with 408 flow (computed as the sum of measured flow in the monthly stress periods. Model boundary conditions Riverside Canal and measured flow at Coffer Dam). were modified at appropriate times during the simulation to represent changes in well pumpage, These computations are shown in figure 20 for drainage of agricultural fields, and channel 1988-92. In general, the model fit is excellent, with the modifications of the Rio Grande. exception of 1990. Of the 348 monthly stress periods The model was calibrated using MODFLOWP not shown in this figure, 346 had an excellent fit and UCODE. Parameter values representing aquifer comparable with those of 1988-89 and 1991-92. The 2 properties and boundary conditions were adjusted years with months of substandard fit were 1975 and through nonlinear regression in a transient-state 1986. The years of substandard fit may have resulted simulation with 96 annual time steps to produce a from Rio Grande flow-measurement errors. model that approximated (1) 4,352 water levels
51 4,000
3,500
Measured Simulated
3,000 y
2,500
2,000
1,500 , in thousands of cubic meters per da Flow
1,000
500
0 v y v Jan Jan Jan Jan Jan Mar Mar Mar Mar Mar July July July July July No No Nov Nov Nov May Ma May May May Sept Sept Sept Sept Sept 1988 1989 1990 1991 1992
Figure 20. Simulated and measured flows at Riverside Dam (location shown in figure 8).
52 measured in 292 wells from 1912 to 1995, (2) three SELECTED REFERENCES seepage-loss rates from a reach of the Rio Grande during periods from 1979 to 1981, (3) three seepage- Anderholm, S.K., and Heywood, C.E., 2003, Chemistry and loss rates from a reach of the Franklin Canal during age of ground water in southwestern Hueco Basin, New Mexico and Texas: U.S. Geological Survey Water- periods from 1990 to 1992, and (4) 24 seepage rates Resources Investigations Report 02-4237. into irrigation drains from 1961 to 1983. Once a Barnes, V.E., 1983, Geologic atlas of Texas, Van Horn - El calibrated model was obtained with MODFLOWP and Paso Sheet: Austin, University of Texas, Bureau of UCODE, the optimal parameter set was used to create Economic Geology, scale 1:250,000. an equivalent MODFLOW-96 simulation with monthly Bennett, G.D., Kontis, A.L., and Larson, S.P., 1982, temporal discretization to improve computations of Representation of multiaquifer well effects in three- seepage from the Rio Grande and to define the flow dimensional ground-water flow simulation: Ground field for a chloride-transport simulation. Water, v. 20, no. 3, p. 334-341. The optimal values for a set of 17 parameters Chapin, C.E., and Seager, W.R., 1975, Evolution of the Rio were obtained using nonlinear regression. Values of Grande rift in the Socorro and Las Cruces areas: New other parameters that were either well constrained or Mexico Geological Society, 26th Field Conference, insensitive to the model were specified. The regression p. 297-321. was constrained by the head and flow-loss Culler, R.C., Hanson, R.L., Myrick, R.M., Turner, R.M., and measurements from the Rio Grande and Franklin Canal Kipple, F.P., 1982, Evapotranspiration before and after clearing phreatophytes, Gila River flood plain, Graham and flow observations in agricultural drains. The model County, Arizona, New Mexico: U.S. Geological Survey was most sensitive to (1) horizontal hydraulic Professional Paper 655-P, 67 p. conductivity of the fluvial facies, which composes the Groschen, G.E., 1994, Simulation of ground-water flow and principal aquifer material in which production wells the movement of saline water in the Hueco Bolson are drilled, (2) specific yield, (3) recharge underflow aquifer, El Paso, Texas, and adjacent areas: U.S. from the Tularosa Basin, and (4) hydraulic Geological Survey Open-File Report 92-171, 87 p. conductance of Quaternary fault zones. Parameter Haneberg, W.C., Mozley, P.S., Moore, J.C., and Goodwin, distribution was generated from a simplified L.B., 1999, Faults and subsurface fluid flow in the hydrogeologic model. Several more complex geologic shallow crust: Geophysical Monograph 113, American conceptualizations, such as a dipping boundary Geophysical Union, Washington, D.C., 222 p. between fluvial and lacustrine-playa facies, were Harbaugh, A.W., and McDonald, M.G., 1996, User’s parameterized and optimized as regression runs. These documentation for MODFLOW-96, an update to the alternative models, though probably more geologically U.S. Geological Survey modular finite-difference realistic, did not provide an overall improved fit to the ground-water flow model: U.S. Geological Survey Open-File Report 96-485, 56 p. hydraulic-head measurements. Future geologic Heywood, C.E., 1993, Monitoring aquifer compaction and refinements incorporated into the flow model may land subsidence due to ground-water withdrawal in the improve model fit in certain areas, however. Seepage El Paso, Texas - Juarez, Chihuahua area, in Prince, losses for the 15-km reach of the Rio Grande channel K.R., Galloway, D.L., and Leake, S.A., eds.: U.S. between the Chamizal zone and Riverside Dam were Geological Survey Subsidence Interest Group computed with monthly stress periods. The seepage Conference, Edwards Air Force Base, Antelope Valley, loss from the Rio Grande for a hypothetical "design Calif., November 18-19, 1992—Abstracts and flow” in the Rio Grande was 39,000 acre-ft/yr and was summary: U.S. Geological Survey Open-File Report not sensitive to specified drain-bed altitudes within a 94-532, 84 p. reasonable range. ———1995, Investigation of aquifer-system compaction in The model input was generated from GIS the Hueco Basin, El Paso, Texas, USA, in Proceedings of the Fifth International Symposium on Land databases, which facilitated rapid model construction Subsidence: International Association of Hydrological and enabled testing of several conceptualizations of Sciences Publication 234, p. 35-45. hydrogeologic facies boundaries. The simulation Hill, M.C., 1992, A computer program (MODFLOWP) for results were sensitive to the hydraulic conductance of estimating parameters of a transient, three-dimensional Quaternary faults in the fluvial-aquifer facies, ground-water flow model using nonlinear regression: suggesting that ground-water flow is impeded across U.S. Geological Survey Open-File Report 91-484, the fault planes. 382 p.
53 Hill, M.C., 1998, Methods and guidelines for effective Proceedings of the National Water Well Association model calibration: U.S. Geological Survey Water- Conference, Practical applications of ground-water Resources Investigations Report 98-4005, 90 p. models, p. 786-796. Holzbecher, E.H., 1998, Modeling density driven flow in McDonald, M.G., and Harbaugh, A.W., 1988, A modular porous media: Berlin, Springer Verlag, 286 p. three-dimensional finite-difference ground-water flow Hsieh, P., 1992, Documentation of a computer program to model: U.S. Geological Survey Techniques of Water- simulate horizontal-flow barriers using the U.S. Resources Investigations, book 6, chap. A1. Geological Survey’s modular three-dimensional finite- McLean, J.S., 1970, Saline ground-water resources of the difference ground-water flow model: U.S. Geological Tularosa Basin, New Mexico: U.S. Department of the Survey Open-File Report 92-477, 32 p. Interior, Office of Saline Water Research and International Boundary and Water Commission Bulletin, Development Progress Report 561, 128 p. 2000. Meyer, W.R., 1976, Digital model for simulated effects of Kernodle, J.M., 1992, Results of simulations by a ground-water pumping in the Hueco Bolson, El Paso preliminary numerical model of land subsidence in the area, Texas, New Mexico, and Mexico: U.S. Geological El Paso, Texas, area: U.S. Geological Survey Water- Survey Water-Resources Investigations 58-75, 31 p. Resources Investigations Report 92-4037, 35 p. Kipp, K.L., Jr., 1987, HST3D—A computer code for Poeter, E.P., and Hill, M.C., 1998, Documentation of simulation of heat and solute transport in three- UCODE, a computer code for universal inverse dimensional ground-water flow systems: U.S. modeling: U.S. Geological Survey Water-Resources Geological Survey Water-Resources Investigations Investigations Report 98-4080, 116 p. Report 86-4095, 517 p. Pollock, D.W., 1994, User’s guide for MODPATH/ Kontis, A.L., and Mandle, R.J., 1988, Modifications of a MODPATH-PLOT, version 3—A particle tracking three-dimensional, ground-water flow model to account post-processing package for MODFLOW, the U.S. for variable water density and effects of multiaquifer Geological Survey finite-difference ground-water flow wells: U.S. Geological Survey Water-Resources model: U.S. Geological Survey Open-File Report Investigations Report 87-4265, 78 p. 94-464, 120 p. Kuniansky, J., and Hillestad, J.G., 1980, Reservoir Prickett, T.A., and Lonnquist, C.G., 1971, Selected digital simulation using bottom-hole pressure boundary computer techniques for groundwater resource conditions: Society of Petroleum Engineers Journal, evaluation: Champaign, Illinois State Water Survey v. 20, p. 473-486. Bulletin 55, 62 p. Land, L.F., and Armstrong, C.A., 1985, A preliminary Prudic, D.E., 1989, Documentation of a computer program assessment of land-surface subsidence in the El Paso to simulate stream-aquifer relations using a modular area, Texas: U.S. Geological Survey Water-Resources finite-difference ground-water flow model: U.S. Investigations Report 85-4155, 96 p. Geological Survey Open-File Report 88-729, 113 p. Langford, R.P., 2001, Segmentation of the Rio Grande Reilly, T.E., Franke, O.L., and Bennett, G.D., 1984, The alluvial surface and evolution of the Rio Grande rift principle of superposition and its application in ground- [abs.]: Proceedings of the 53d Annual Meeting (April water hydraulics: U.S. Geological Survey Open-File 29–May 2, 2001) of the Rocky Mountain Section, Report 84-459, 36 p. Geological Society of America, Boulder, Colo. Riley, F.S., 1998, Mechanics of aquifer systems—The Lee Wilson and Associates, 1985a, Report scientific legacy of Joseph F. Poland, in Borchers, J.W., 3—Hydrogeology of the Hueco Basin: Prepared for the ed., Land subsidence—Case studies and current Public Services Board, City of El Paso, Texas. research: Proceedings of the Dr. Joseph F. Poland ———1985b, Report 4—Technical framework for Symposium on Land Subsidence, Association of evaluation of proposed Hueco Basin appropriations: Engineering Geologists Special Publication no. 8, Prepared for the Public Services Board, City of El Paso, p. 13-27. Texas. ———1991, Research Memoranda 3—Solute transport Robinson, T.W., 1958, Phreatophytes: U.S. Geological model for the Hueco Bolson Recharge Project: Survey Water-Supply Paper 1423, 84 p. Prepared for the Public Services Board, City of El Paso, Sayre, A.N., and Livingston, P., 1945, Ground-water Texas. resources of the El Paso area, Texas: U.S. Geological Leggat, E.R., and Davis, M.E., 1966, Analog model study of Survey Water-Supply Paper 919, 190 p. the Hueco Bolson near El Paso, Texas: Texas Water Seager, W.R., 1982, Geology of Organ Mountains and Development Board Report 28, 26 p. southern San Andres Mountains, New Mexico: McDonald, M.G., 1984, Development of a multi-aquifer Socorro, New Mexico Bureau of Mines and Mineral well option for a modular ground-water flow model: Resources Memoir 36.
54 Seager, W.R., Hawley, J.W., Kottlowski, F.E., and Kelley, S.A., 1987, Geology of the east half of Las Cruces and northeast El Paso 1 x 2 sheets (scale 1:125,000), New Mexico: Socorro, New Mexico Bureau of Mines and Mineral Resources, 5 sheets. Slichter, C.S., 1905, Observations on the ground waters of the Rio Grande Valley: U.S. Geological Survey Water- Supply Paper 141, 83 p. Strain, W.S., 1969, Late Cenozoic strata of the El Paso area, in Kottlowski, F.E., and Lemone, D.V., eds., Border Stratigraphy Symposium: Socorro, New Mexico Bureau of Mines and Mineral Resources Circular 104, 123 p. Thiem, G., 1906, Hydrologische methoden: Gebhardt, Leipzig, 56 p. Trescott, P.C., and Larson, S.P., 1976, Finite-difference model for aquifer simulation in two dimensions with results of numerical experiments: U.S. Geological Survey Techniques of Water-Resources Investigations, book 6, chap. A1, 116 p. U.S. Department of Commerce, U.S. Census Bureau, 2002, accessed February 26, 2002, at http://www.census.gov/ prod/2002pubs/01statab/pop.pdf. Waltemeyer, S.D., 2001, Estimates of mountain-front streamflow available for potential recharge to the Tularosa Basin, New Mexico: U.S. Geological Survey Water-Resources Investigations Report 01-4013, 8 p. Weeks, E.P., Weaver, H.L., Campbell, G.S., and Tanner, B.D., 1987, Water use by salt cedar and by replacement vegetation in the Pecos River flood plain between Acme and Artesia, New Mexico: U.S. Geological Survey Professional Paper 491-G, 33 p. White, D.E., Baker, E.T., and Sperka, Roger, 1997, Hydrology of the shallow aquifer and uppermost semiconfined aquifer near El Paso, Texas: U.S. Geological Survey Water-Resources Investigations Report 97-4263, 37 p. Wilkins, D.W., 1998, Summary of the Southwest Alluvial Basins Regional Aquifer-System Analysis in parts of Colorado, New Mexico, and Texas: U.S. Geological Survey Professional Paper 1407-A, 49 p. Zheng, Chunmiao, and Wang, P.P., 1998, MT3DMS—A modular three-dimensional multispecies transport model for simulation of advection, dispersion, and chemical reactions of contaminants in groundwater systems, Documentation and user’s guide, contract report SERD P-99-1: U.S. Army Engineer Research and Development Center, Vicksburg, Miss.
55 APPENDIX 1: MODIFICATIONS TO MODFLOWP
Changes made to modflowp.f for all modifications DATA CUNIT/'BCF ', 'WEL ', 'DRN ', 'RIV ', 'EVT ', 'TLK ', 'GHB ', & 'RCH ', 'SIP ', 'DE4 ', 'SOR ', 'OC ', 'PCG ', 'GFD ', & 'PAR ', 'HFB ', 'RES ', 'STR ', 'IBS ', 'CHD ', 'MAW ', & ' ', ' ', ' ', ' ', ' ', ' ', ' ', & ' ', ' ', ' ', ' ', ' ', ' ', ' ', & ' ', ' ', ' ', ' ', ' '/ ... skip lines... IF (IUNIT(2).GT.0) CALL WEL5AL(ISUM,LENX,LCWELL,MXWELL,NWELLS, & IUNIT(2),IOUT,IWELCB,NWELVL,IWELAL, & IFREFM) IF(IUNIT(21).GT.0) CALL MAW5AL(ISUM,LENX,MXMAW,NMAWS, 1 IUNIT(21),IOUT,NLAY,LCRMAW,LCRMAL,IMAWCB) ... skip lines... IF (IUNIT(18).GT.0) CALL STR5AL(ISUM,LENX,LCSTRM,ICSTRM,MXSTRM, & NSTREM,IUNIT(18),IOUT,ISTCB1, & ISTCB2,NSS,NTRIB,NDIV,ICALC,CONST, & LCTBAR,LCTRIB,LCIVAR,LCFGAR,NSTROP) ... skip lines... IF (IUNIT(2).GT.0) CALL WEL5RP(X(LCWELL),NWELLS,MXWELL, & IUNIT(2),IOUT,NWELVL,IWELAL, & IFREFM) IF(IUNIT(21).GT.0) CALL MAW5RP(X(LCRMAW), 1 X(LCRMAL),NMAWS,MXMAW,IUNIT(21),IOUT,NLAY) ... skip lines... IF (IUNIT(2).GT.0) CALL WEL5FM(NWELLS,MXWELL,X(LCRHS), & X(LCWELL),X(LCIBOU),NCOL, & NROW,NLAY,NWELVL) IF(IUNIT(21).GT.0) CALL MAW5FM(NMAWS,MXMAW,X(LCRHS),X(LCHCOF), 1 X(LCIBOU),X(LCRMAW),X(LCRMAL),NCOL,NROW, 2 NLAY,X(LCCR),X(LCDELC),X(LCDELR),X(LCHNEW),IOUT) ... skip lines... IF (IUNIT(18).GT.0) CALL STR5FM(NSTREM,X(LCSTRM),X(ICSTRM), & X(LCHNEW),X(LCHCOF),X(LCRHS) & ,X(LCIBOU),MXSTRM,NCOL,NROW, & NLAY,IOUT,NSS,X(LCTBAR), & NTRIB,X(LCTRIB),X(LCIVAR), & X(LCFGAR),ICALC,CONST,NSTROP, & X(LCSHNW),IUNIT(15),IP,ISN) ... skip lines... IF (IUNIT(2).GT.0) CALL WEL5BD(NWELLS,MXWELL,VBNM,VBVL,MSUM, & X(LCWELL),X(LCIBOU),DELT, & NCOL,NROW,NLAY,KKSTP,KKPER, & IWELCB,ICBCFL,X(LCBUFF),IOUT, & PERTIM,TOTIM,NWELVL,IWELAL) IF(IUNIT(21).GT.0) CALL MAW5BD(NMAWS,MXMAW,X(LCIBOU),X(LCRMAW), 1 X(LCRMAL),NCOL,NROW,NLAY,X(LCHNEW),KSTP, 2 KPER,IMAWCB,ICBCFL,X(LCBUFF),IOUT,MSUM,DELT,VBNM,VBVL) ... skip lines...
56 IF (IUNIT(18).GT.0) CALL STR5BD(NSTREM,X(LCSTRM),X(ICSTRM), & X(LCIBOU),MXSTRM,X(LCHNEW), & NCOL,NROW,NLAY,DELT,VBVL, & VBNM,MSUM,KKSTP,KKPER, & ISTCB1,ISTCB2,ICBCFL, & X(LCBUFF),IOUT,NTRIB,NSS, & X(LCTRIB),X(LCTBAR), & X(LCIVAR),X(LCFGAR),ICALC, & CONST,IPTFLG,NSTROP) ... skip lines... CALL SEN1OT(IUHEAD,IOUT,NROW,NCOL,NLAY,NP,ISN,NH,X(LCNDER), & X(LCHNEW),PID,DID,IP,KPER,X(LCBUFF),KSTP,PERTIM,TOTIM, & X(LCX),X(LCDELR),X(LCDELC),X(LCIBOU),X(LCCOFF), & X(LCROFF),X(LCH),X(LCWT),X(LCHOBS),IPRINT,IFO,ITERP, & IPAR,X(LCRINT),X(LCJOFF),X(LCIOFF),X(LCMLAY),X(LCPR), & MOBS,NPER,X(LCB),X(LCLN),NQ,NQC,NQT,X(LCNQOB), & X(LCNQCL),X(LCIQOB),X(LCQCLS),X(LCIBT),MXBND,NBOUND, & X(LCBNDS),MXRIVR,NRIVER,X(LCRIVR),X(LCSHNW),LASTX, & ISCALS,X(LCTOFF),MXDRN,NDRAIN,X(LCDRAI),MXSTRM,NSTREM, & X(LCSTRM),X(ICSTRM),MAXM,KPRINT,JDRY,IDRY,NPR,X(LCWP), & MPR,X(LCPRM),X(LCIWPG),X(LCB1),IOUB,RSQ,RSQP,RSQO, & RSQOO,NPO,SOSC,SOSR,IPR,X(LCNIPR),X(LCWPF),ND,RSQF, & IOUYR,IOUHDS,IOUFLW,IOUPRI,IUNORM,X(LCWTQ),X(LCWTQS), & IOWTQ,NDMH,X(LCPV),X(LCRHS),X(LCCR),X(LCCC),X(LCCV), & X(LCBOT),X(LCTOP),NPTH,X(LCNPNT),NTT2,KTDIM,KTREV, & X(LCICLS),X(LCPRST),X(LCPOFF),X(LCSMAT),X(LCNLL),NSM, & X(LCTRPY),NMM,NZM,LZI1,X(LCSFAC),X(LCLZ),X(LCLM), & X(LCMATZ),NLLI1,X(LCST),X(LCTT2),IOUTT2,NRCHOP, & X(LCIRCH),X(LCRECH),X(LCSV),X(LCBANI),X(LCTHCK),NCLAY, & KTFLG,ADVSTP,X(LCLAYC),NSTROP)
57 Compute heads for multi-layer observations with dry cells: ssen1jz.f
Subroutine SSEN1U This subroutine interpolates heads and accounts for dry cells, if necessary. REAL B, COFF, DELC, DELR, FACT, H, HD, HOBS, PR, PROP, RHS, RINT, & ROFF, TOFF, W, WT, X, ZERO, PROLD ...skip lines... C------IF THE OBSERVATION THIS IS TO BE SUBTRACTED FROM IS DRY, MAKE C------THIS ONE DRY, TOO N1 = NDER(5,N) IF (N1.GT.0) THEN IF ((WT(N1).LT.ZERO.OR.COFF(N1).GE.5.)) THEN IDRY = IDRY + 1 WT(N) = -ABS(WT(N)) WRITE (IOUT,500) N, DID(N) GOTO 30 ENDIF ENDIF C------CHECK FOR DRY OBSERVATIONS OR INTERPOLATIONS AFFECTED BY DRY C------CELLS DO 20 M = 1, MM KK = K IF (K.LT.0) KK = MLAY(M,ML) IF (KK.EQ.0) GOTO 30 IF (LAYCON(KK).EQ.1 .OR. LAYCON(KK).EQ.3) THEN IF (IBOUND(JJ,II,KK).EQ.0) THEN C*** dry cell in observation stack: C*** more than 1 layer in observations C*** change MLAY (move layer definitions forward in array) C*** recalculate pr C*** at least 1 non dry layer? C*** goto 20 IF (MLAY(M+1,ML).eq.0) GOTO 13 PROLD = PR(M,ML) DO 11 MTMP = M, MM-1 IF (MLAY(MTMP,ML).eq.0) GOTO 12 MLAY(MTMP,ML) = MLAY(MTMP+1,ML) PR(MTMP,ML) = PR(MTMP+1,ML)/(1-PROLD) IF (PR(MTMP+1,ML).eq.0) PR(MTMP,ML) = 0 11 CONTINUE MLAY(MM,ML) = 0 PR(MM,ML) = 0 12 IF (MLAY(1,ML).gt.0) GOTO 20 13 IDRY = IDRY + 1 WT(N) = -ABS(WT(N)) WRITE (IOUT,500) N, DID(N) GOTO 30 ELSEIF ((RINT(2,N).NE.ZERO.AND.IBOUND(JJ+JO,II,KK) & .EQ.0) .OR. & (RINT(3,N).NE.ZERO.AND.IBOUND(JJ,II+IO,KK) & .EQ.0) .OR. & (RINT(4,N).NE.ZERO.AND.IBOUND(JJ+JO,II+IO,KK) & .EQ.0)) THEN C*** dry cell in adjacent stack: C*** IF (MM.GT.1 .OR. TOFF(N).GT.ZERO) THEN C*** multi-layer observation IF (MM.GT.1) THEN C*** adjust PR and ML arrays as above IF (MLAY(M+1,ML).eq.0) GOTO 16 PROLD = PR(M,ML) DO 14 MTMP = M, MM-1
58 IF (MLAY(MTMP,ML).eq.0) GOTO 15 MLAY(MTMP,ML) = MLAY(MTMP+1,ML) PR(MTMP,ML) = PR(MTMP+1,ML)/(1-PROLD) IF (PR(MTMP+1,ML).eq.0) PR(MTMP,ML) = 0 14 CONTINUE MLAY(MM,ML) = 0 PR(MM,ML) = 0 15 IF (MLAY(1,ML).gt.0) GOTO 20 ENDIF 16 IF (MM.GT.1 .OR. TOFF(N).GT.ZERO) THEN IDRY = IDRY + 1 WT(N) = -ABS(WT(N)) WRITE (IOUT,500) N, DID(N) GOTO 30 ENDIF
Compute leakage from streams to underlying active cells: str5p.f
Subroutine STR5AL This subroutine allocates array storage for streams. SUBROUTINE STR5AL(ISUM,LENX,LCSTRM,ICSTRM,MXSTRM,NSTREM,IN,IOUT, & ISTCB1,ISTCB2,NSS,NTRIB,NDIV,ICALC,CONST,LCTBAR, & LCTRIB,LCIVAR,LCFGAR,NSTROP) ... skip lines... READ (IN,505) MXSTRM, NSS, NTRIB, NDIV, ICALC, CONST, ISTCB1, & ISTCB2,NSTROP 505 FORMAT (5I10,F10.0,3I10) ... skip lines... 540 FORMAT (' ***X ARRAY MUST BE DIMENSIONED LARGER***') C C3------CHECK TO SEE THAT OPTION IS LEGAL. IF(NSTROP.GE.1.AND.NSTROP.LE.3) GO TO 560 C C3A-----IF ILLEGAL PRINT A MESSAGE AND ABORT SIMULATION WRITE(IOUT,550) 550 FORMAT(1X,'ILLEGAL OPTION CODE. SIMULATION ABORTING') STOP C C4------PRINT OPTION CODE. 560 IF(NSTROP.EQ.1) WRITE(IOUT,565) 565 FORMAT(1X,'OPTION 1 -- LEAKAGE TO DEFINED LAYER') IF(NSTROP.EQ.3) WRITE(IOUT,570) 570 FORMAT(1X,'OPTION 3 -- LEAKAGE TO HIGHEST ACTIVE NODE IN EACH', 1 ' VERTICAL COLUMN') C C10-----RETURN. RETURN END
Subroutine STR5FM This subroutine adds stream terms to RHS and HCOF if flow occurs in model cell. SUBROUTINE STR5FM(NSTREM,STRM,ISTRM,HNEW,HCOF,RHS,IBOUND,MXSTRM, & NCOL,NROW,NLAY,IOUT,NSS,ITRBAR,NTRIB,ARTRIB, & IDIVAR,NDFGAR,ICALC,CONST,NSTROP,SHNW,IU,IP,ISN) ... skip lines ...
59 C12----DETERMINE LEAKAGE THROUGH STREAMBED. C IF DRY CELL AND OPTION 3, CHECK FOR HIGHEST ACTIVE CELL IF ((IBOUND(IC,IR,IL).LT.1).AND.(NSTROP.EQ.1)) THEN FLOBOT = ZERO ELSE IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.3)) THEN C12A------IF OPTION IS 3 LEAKAGE IS INTO HIGHEST INTERNAL CELL. C CANNOT PASS THROUGH CONSTANT HEAD NODE DO 42 NIL=IL+1,NLAY C C12B-----IF CELL IS: CONSTANT HEAD MOVE ON TO NEXT STREAM REACH C INACTIVE MOVE DOWN A LAYER C ACTIVE SET LEAKAGE TO CONSTANT IF(IBOUND(IC,IR,NIL)) 43,42,44 42 CONTINUE C------NO ACTIVE CELLS FOUND, NO LEAKAGE 43 FLOBOT=0. GO TO 45 ENDIF 44 IF (FLOWIN.LE.ZERO) HSTR = STRM(5,L) CSTR = STRM(3,L) SBOT = STRM(4,L) H = HNEW(IC,IR,IL) C*** set head to top active layer if dry cell IF(IBOUND(IC,IR,IL).EQ.0) H=HNEW(IC,IR,NIL) C-----ADDED FOR PARAMETER ESTIMATION IF (IU.GT.0 .AND. IP.NE.0) H = SHNW(IC,IR,IL) IF(IU.GT.0.AND.IP.NE.0.AND.IBOUND(IC,IR,IL).EQ.0) + H=SHNW(IC,IR,NIL) ... skip lines... C16-----STREAMFLOW OUT EQUALS STREAMFLOW IN MINUS LEAKAGE. IF ((IBOUND(IC,IR,IL).LT.1).AND.(NSTROP.EQ.1)) FLOBOT = ZERO ENDIF ENDIF 45 FLOWOT = FLOWIN - FLOBOT IF (ISTSG.GT.1 .AND. NREACH.EQ.1) STRM(9,LL) = ARTRIB(IFLG) C C17----STORE STREAM INFLOW, OUTFLOW AND LEAKAGE FOR EACH REACH. STRM(9,L) = FLOWOT STRM(10,L) = FLOWIN STRM(11,L) = FLOBOT C C18----RETURN TO STEP 3 IF STREAM INFLOW IS LESS THAN OR EQUAL TO ZERO C AND LEAKAGE IS GREATER THAN OR EQUAL TO ZERO OR IF CELL C IS NOT ACTIVE--IBOUND IS LESS THAN OR EQUAL TO ZERO-- C & NSTROP=1. IF ((IBOUND(IC,IR,IL).GT.0).OR.(NSTROP.EQ.3)) THEN C*** reset top active layer if dry cell IF(IBOUND(IC,IR,IL).EQ.0) IL=NIL IF (FLOWIN.GT.ZERO .OR. FLOBOT.LT.ZERO) THEN C C19------IF HEAD > BOTTOM THEN ADD TERMS TO RHS AND HCOF. IF (IQFLG.LT.1) THEN C------FOR ADJOINT STATES, ONLY CALCULATE CONTRIBUTION TO HCOF IF (IU.EQ.0 .OR. IP.EQ.0 .OR. ISN.LT.0) & RHS(IC,IR,IL) = RHS(IC,IR,IL) - CSTR*HSTR HCOF(IC,IR,IL) = HCOF(IC,IR,IL) - CSTR ELSE C C20------IF HEAD < BOTTOM THEN ADD TERM ONLY TO RHS. C------FOR ADJOINT STATES, ONLY CALCULATE CONTRIBUTION TO HCOF IF (IU.EQ.0 .OR. IP.EQ.0 .OR. ISN.LT.0) THEN RHS(IC,IR,IL)=RHS(IC,IR,IL) - FLOBOT
60 ENDIF ENDIF ENDIF ENDIF 50 CONTINUE
Subroutine STR5BD This subroutine calculates volumetric budget for streams. SUBROUTINE STR5BD(NSTREM,STRM,ISTRM,IBOUND,MXSTRM,HNEW,NCOL,NROW, & NLAY,DELT,VBVL,VBNM,MSUM,KSTP,KPER,ISTCB1, & ISTCB2,ICBCFL,BUFF,IOUT,NTRIB,NSS,ARTRIB,ITRBAR, & IDIVAR,NDFGAR,ICALC,CONST,IPTFLG,NSTROP) ... skip lines... C14----DETERMINE LEAKAGE THROUGH STREAMBED. IF ((IBOUND(IC,IR,IL).LT.1).AND.(NSTROP.EQ.1)) THEN FLOBOT = ZERO ELSE IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.3)) THEN DO 72 NIL=IL+1,NLAY C C18A-----IF CELL IS: CONSTANT HEAD MOVE ON TO NEXT STREAM REACH C INACTIVE MOVE DOWN A LAYER C ACTIVE SET LEAKAGE TO CONSTANT IF(IBOUND(IC,IR,NIL)) 73,72,75 72 CONTINUE C------NO ACTIVE CELLS FOUND, NO LEAKAGE 73 FLOBOT=0. GO TO 77 75 IF (STRM(3,L).ne.0) & WRITE (IOUT,400) NIL, ISTRM(4,L), ISTRM(5,L), IR, IC 400 FORMAT (/,5X,'LEAKAGE TO LAYER',I5,' FROM STREAM SEG', & I6,' STREAM REACH',I6,' IN ROW',I5,' COLUMN',I5) ENDIF IF (FLOWIN.LE.ZERO) HSTR = STRM(5,L) CSTR = STRM(3,L) SBOT = STRM(4,L) H = HNEW(IC,IR,IL) C*** set head to top active layer if dry cell IF(IBOUND(IC,IR,IL).EQ.0) H=HNEW(IC,IR,NIL) ... skip lines... C18----STREAMFLOW OUT EQUALS STREAMFLOW IN MINUS LEAKAGE. IF ((IBOUND(IC,IR,IL).LT.1).AND.(NSTROP.EQ.1)) & FLOBOT = ZERO ENDIF ENDIF 77 FLOWOT = FLOWIN - FLOBOT IF (ISTSG.GT.1 .AND. NREACH.EQ.1) STRM(9,LL) = ARTRIB(IFLG) C C19----STORE STREAM INFLOW, OUTFLOW AND LEAKAGE FOR EACH REACH. STRM(9,L) = FLOWOT STRM(10,L) = FLOWIN STRM(11,L) = FLOBOT C C20----IF LEAKAGE FROM STREAMS IS TO BE SAVED THEN ADD RATE TO BUFFER. C------OPTION=3; LEAKAGE IS INTO HIGHEST CELL IN A VERTICAL COLUMN C------THAT IS NOT NO FLOW. IF NO ACTIVE CELLS EXIST THEN ZERO STREAM C------LEAKAGE IF ((IBD.EQ.1).AND.(FLOBOT.NE.0.0)) THEN IF(IBOUND(IC,IR,IL).GT.0) THEN
61 BUFF(IC,IR,IL)=BUFF(IC,IR,IL) + FLOBOT ELSE BUFF(IC,IR,NIL)=BUFF(IC,IR,NIL) + FLOBOT ENDIF ENDIF C C21----DETERMINE IF FLOW IS INTO OR OUT OF MODEL CELL. ... skip lines... C29-----SAVE STREAMFLOWS OUT OF EACH REACH ON DISK. DO 120 L = 1, NSTREM IC = ISTRM(3,L) IR = ISTRM(2,L) IL = ISTRM(1,L) IF (IBOUND(IC,IR,IL).GT.0) & BUFF(IC,IR,IL) = BUFF(IC,IR,IL) + STRM(9,L) C*** reset top active layer if dry cell IF ((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.3)) THEN DO 118 NIL=IL+1,NLAY C C29A-----IF CELL IS: CONSTANT HEAD MOVE ON TO NEXT STREAM REACH C INACTIVE MOVE DOWN A LAYER IF(IBOUND(IC,IR,NIL)) 118,119,118 118 CONTINUE GO TO 120 119 BUFF(IC,IR,NIL) = BUFF(IC,IR,NIL) + STRM(9,L) ENDIF 120 CONTINUE CALL UBUDSV(KSTP,KPER,STRTXT,ISTCB2,BUFF,NCOL,NROW,NLAY,IOUT) Subroutine PAR1AQ This subroutine calculates final iteration parameters and updates parameters. CALL SSEN1V(NQ,NQC,NQT,NQOB,NQCL,IQOB,QCLS,IBT,MXBND,NBOUND, & BNDS,MXRIVR,NRIVER,RIVR,SHNW,IP,HNEW,NCOL,NROW,NLAY, & IOUT,IBOUND,NPER,KPER,NH,X,DID,NP,H,B,LN,TOFF, & MXDRN,NDRAIN,DRAI,MXSTRM,NSTREM,STRM,ISTRM,ISN, & WTQ,NDMH,NSTROP)
Subroutine SEN1OT This subroutine prints data for observed heads and flows. SUBROUTINE SEN1OT(IUHEAD,IOUT,NROW,NCOL,NLAY,NP,ISN,NH,NDER,HNEW, & PID,DID,IP,KPER,BUFF,KSTP,PERTIM,TOTIM,X,DELR, & DELC,IBOUND,COFF,ROFF,H,WT,HOBS,IPRINT,IFO, & ITERP,IPAR,RINT,JOFF,IOFF,MLAY,PR,MOBS,NPER,B, & LN,NQ,NQC,NQT,NQOB,NQCL,IQOB,QCLS,IBT,MXBND, & NBOUND,BNDS,MXRIVR,NRIVER,RIVR,SHNW,LASTX, & ISCALS,TOFF,MXDRN,NDRAIN,DRAI,MXSTRM,NSTREM, & STRM,ISTRM,MAXM,KPRINT,JDRY,IDRY,NPR,WP,MPR,PRM, & IWPG,B1,IOUB,RSQ,RSQP,RSQO,RSQOO,NPO,SOSC,SOSR, & IPR,NIPR,WPF,ND,RSQF,IOUYR,IOUHDS,IOUFLW,IOUPRI, & IUNORM,WTQ,WTQS,IOWTQ,NDMH,PV,RHS,CR,CC,CV,BOT, & TOP,NPTH,NPNT,NTT2,KTDIM,KTREV,ICLS,PRST,POFF, & SMAT,NLL,NSM,TRPY,NMM,NZM,LZI1,SFAC,LZ,LM,MATZ, & NLLI1,ST,TT2,IOUTT2,NRCHOP,IRCH,RECH,SV,BANIV, & THCK,NCLAY,KTFLG,ADVSTP,LAYC,NSTROP)
... skip lines... IF (NQ.GT.0) CALL SSEN1V(NQ,NQC,NQT,NQOB,NQCL,IQOB,QCLS,IBT,
62 & MXBND,NBOUND,BNDS,MXRIVR,NRIVER,RIVR, & SHNW,IP,HNEW,NCOL,NROW,NLAY,IOUT, & IBOUND,NPER,KPER,NH,X,DID,NP,H,B,LN, & TOFF,MXDRN,NDRAIN,DRAI,MXSTRM,NSTREM, & STRM,ISTRM,ISN,WTQ,NDMH,NSTROP)
Subroutine SSEN1V This subroutine saves simulated flows and calculates sensitivities. CHANGE 28.02.96:ARGUMENTS ADDED SUBROUTINE AND DIMENSION STATEMENT SUBROUTINE SSEN1V(NQ,NQC,NQT,NQOB,NQCL,IQOB,QCLS,IBT,MXBND, & NBOUND,BNDS,MXRIVR,NRIVER,RIVR,SHNW,IP,HNEW,NCOL,NROW, & NLAY,IOUT,IBOUND,NPER,KPER,NH,X,DID,NP,H,B,LN, & TOFF,MXDRN,NDRAIN,DRAI,MXSTRM,NSTREM,STRM,ISTRM, & ISN,WTQ,NDMH,NSTROP) ... skip lines... C------ASSIGN VARIABLE VALUES IF (IBT1.EQ.3) THEN K = ISTRM(1,NB) I = ISTRM(2,NB) J = ISTRM(3,NB) C IF DRY CELL AND OPTION 3, CHECK FOR HIGHEST ACTIVE CELL IF ((IBOUND(J,I,K).LT.1).AND.(NSTROP.EQ.1)) GOTO 30 IF ((IBOUND(J,I,K).LE.0).AND.(NSTROP.EQ.3)) THEN C------IF OPTION IS 3 LEAKAGE IS INTO HIGHEST INTERNAL CELL. C CANNOT PASS THROUGH CONSTANT HEAD NODE DO 5 NIL=K+1,NLAY C C-----IF CELL IS: CONSTANT HEAD MOVE ON TO NEXT STREAM REACH C INACTIVE MOVE DOWN A LAYER C ACTIVE SET LEAKAGE TO CONSTANT IF(IBOUND(J,I,NIL)) 6,5,7 5 CONTINUE C------NO ACTIVE CELLS FOUND, NO LEAKAGE 6 GO TO 30 ENDIF ENDIF 7 IF (IP.EQ.0) HHNEW = HNEW(J,I,K) IF (IP.GT.0) HHNEW = SHNW(J,I,K)
APPENDIX 2: MODIFICATIONS TO MODFLOW
Changes made to modflw96.f for all modifications DATA CUNIT/'BCF ','WEL ','DRN ','RIV ','EVT ','TLK ','GHB ', 1 'RCH ','SIP ','DE4 ','SOR ','OC ','PCG ','GFD ', 2 'MAW ','HFB ','RES ','STR ','IBS ','CHD ','FHB ', 3 ' ',' ',' ',' ',' ',' ',' ', 4 ' ',' ',' ',' ',' ',' ',' ', 5 ' ',' ',' ',' ',' '/ ... skip lines... IF(IUNIT(2).GT.0) CALL WEL5AL(ISUM,LENX,LCWELL,MXWELL,NWELLS, 1 IUNIT(2),IOUT,IWELCB,NWELVL,IWELAL,IFREFM) C*** multi-layer wells IF(IUNIT(15).GT.0) CALL MAW5AL(ISUM,LENX,MXMAW,NMAWS,
63 1 IUNIT(15),IOUT,NLAY,LCRMAW,LCRMAL,IMAWCB) ... skip lines... IF(IUNIT(18).GT.0) CALL STR1AL(ISUM,LENX,LCSTRM,ICSTRM,MXSTRM, 1 NSTREM,IUNIT(18),IOUT,ISTCB1,ISTCB2,NSS,NTRIB, 2 NDIV,ICALC,CONST,LCTBAR,LCTRIB,LCIVAR,LCFGAR, 3 NSTROP) ... skip lines... IF(IUNIT(2).GT.0) CALL WEL5RP(X(LCWELL),NWELLS,MXWELL,IUNIT(2), 1 IOUT,NWELVL,IWELAL,IFREFM) C*** multi-layer wells IF(IUNIT(15).GT.0) CALL MAW5RP(X(LCRMAW), 1 X(LCRMAL),NMAWS,MXMAW,IUNIT(15),IOUT,NLAY) ... skip lines... IF(IUNIT(2).GT.0) CALL WEL5FM(NWELLS,MXWELL,X(LCRHS),X(LCWELL), 1 X(LCIBOU),NCOL,NROW,NLAY,NWELVL) C*** multi-layer wells IF(IUNIT(15).GT.0) CALL MAW5FM(NMAWS,MXMAW,X(LCRHS),X(LCHCOF), 1 X(LCIBOU),X(LCRMAW),X(LCRMAL),NCOL,NROW, 2 NLAY,X(LCCR),X(LCDELC),X(LCDELR),X(LCHNEW),IOUT) ... skip lines... IF(IUNIT(18).GT.0) CALL STR1FM(NSTREM,X(LCSTRM),X(ICSTRM), 1 X(LCHNEW),X(LCHCOF),X(LCRHS),X(LCIBOU), 2 MXSTRM,NCOL,NROW,NLAY,IOUT,NSS,X(LCTBAR), 3 NTRIB,X(LCTRIB),X(LCIVAR),X(LCFGAR),ICALC,CONST, 4 NSTROP) ... skip lines... IF(IUNIT(2).GT.0) CALL WEL5BD(NWELLS,MXWELL,VBNM,VBVL,MSUM, 1 X(LCWELL),X(LCIBOU),DELT,NCOL,NROW,NLAY,KKSTP,KKPER,IWELCB, 1 ICBCFL,X(LCBUFF),IOUT,PERTIM,TOTIM,NWELVL,IWELAL) C*** multi-layer wells IF(IUNIT(15).GT.0) CALL MAW5BD(NMAWS,MXMAW,X(LCIBOU),X(LCRMAW), 1 X(LCRMAL),NCOL,NROW,NLAY,X(LCHNEW),KSTP, 2 KPER,IMAWCB,ICBCFL,X(LCBUFF),IOUT,MSUM,DELT,VBNM,VBVL) ... skip lines... IF(IUNIT(18).GT.0) CALL STR1BD(NSTREM,X(LCSTRM),X(ICSTRM), STR1 1 X(LCIBOU),MXSTRM,X(LCHNEW),NCOL,NROW,NLAY,DELT,VBVL,VBNM,MSUM, STR1 2 KKSTP,KKPER,ISTCB1,ISTCB2,ICBCFL,X(LCBUFF),IOUT,NTRIB,NSS, STR1 3 X(LCTRIB),X(LCTBAR),X(LCIVAR),X(LCFGAR),ICALC,CONST,IPTFLG, 4 NSTROP)
Compute leakage from streams to underlying active cells: str1.f
Subroutine STR1AL This subroutine allocates array storage for streams. SUBROUTINE STR1AL(ISUM,LENX,LCSTRM,ICSTRM,MXSTRM,NSTREM,IN, 1 IOUT,ISTCB1,ISTCB2,NSS,NTRIB,NDIV,ICALC,CONST, 2 LCTBAR,LCTRIB,LCIVAR,LCFGAR,NSTROP) ... skip lines... C2------READ MXSTRM, NSS, NTRIB, ISTCB1, AND ISTCB2. 100 READ(IN,3)MXSTRM,NSS,NTRIB,NDIV,ICALC,CONST,ISTCB1,ISTCB2,NSTROP
64 3 FORMAT(5I10,F10.0,3I10) ... skip lines... 10 FORMAT(1X,' ***X ARRAY MUST BE DIMENSIONED LARGER***') C C3------CHECK TO SEE THAT OPTION IS LEGAL. IF(NSTROP.GE.1.AND.NSTROP.LE.3) GO TO 250 C C3A-----IF ILLEGAL PRINT A MESSAGE AND ABORT SIMULATION WRITE(IOUT,11) 11 FORMAT(1X,'ILLEGAL OPTION CODE. SIMULATION ABORTING') STOP C C4------PRINT OPTION CODE. 250 IF(NSTROP.EQ.1) WRITE(IOUT,12) 12 FORMAT(1X,'OPTION 1 -- LEAKAGE TO DEFINED LAYER') IF(NSTROP.EQ.3) WRITE(IOUT,13) 13 FORMAT(1X,'OPTION 3 -- LEAKAGE TO HIGHEST ACTIVE NODE IN EACH', 1 ' VERTICAL COLUMN') C10-----RETURN. RETURN END
Subroutine STR1FM This subroutine adds stream terms to RHS and HCOF if flow occurs in model cell. SUBROUTINE STR1FM(NSTREM,STRM,ISTRM,HNEW,HCOF,RHS,IBOUND,MXSTRM, 1 NCOL,NROW,NLAY,IOUT,NSS,ITRBAR,NTRIB,ARTRIB, 2 IDIVAR,NDFGAR,ICALC,CONST,NSTROP) ... skip lines... C12----DETERMINE LEAKAGE THROUGH STREAMBED. C C IF DRY CELL AND OPTION 3, CHECK FOR HIGHEST ACTIVE CELL IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.1)) GO TO 315 IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.3)) THEN C12A------IF OPTION IS 3 LEAKAGE IS INTO HIGHEST INTERNAL CELL. C CANNOT PASS THROUGH CONSTANT HEAD NODE DO 3101 NIL=IL+1,NLAY C C12B-----IF CELL IS: CONSTANT HEAD MOVE ON TO NEXT STREAM REACH C INACTIVE MOVE DOWN A LAYER C ACTIVE SET LEAKAGE TO CONSTANT IF(IBOUND(IC,IR,NIL)) 3102,3101,311 3101 CONTINUE C------NO ACTIVE CELLS FOUND, NO LEAKAGE 3102 FLOBOT=0. GO TO 320 ENDIF 311 IF(FLOWIN.LE.0.) HSTR=STRM(5,L) CSTR=STRM(3,L) SBOT=STRM(4,L) H=HNEW(IC,IR,IL) C*** set head to top active layer if dry cell IF(IBOUND(IC,IR,IL).EQ.0) H=HNEW(IC,IR,NIL) T=HSTR-SBOT ... skip lines... C16-----STREAMFLOW OUT EQUALS STREAMFLOW IN MINUS LEAKAGE. 315 IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.1)) FLOBOT=0. 320 FLOWOT=FLOWIN-FLOBOT
65 IF((ISTSG.GT.1).AND.(NREACH.EQ.1)) STRM(9,LL)=ARTRIB(IFLG) C C17----STORE STREAM INFLOW, OUTFLOW AND LEAKAGE FOR EACH REACH. STRM(9,L)=FLOWOT STRM(10,L)=FLOWIN STRM(11,L)=FLOBOT C C18----RETURN TO STEP 3 IF STREAM INFLOW IS LESS THAN OR EQUAL TO ZERO C AND LEAKAGE IS GREATER THAN OR EQUAL TO ZERO, OR IF CELL C IS NOT ACTIVE--IBOUND IS LESS THAN OR EQUAL TO ZERO-- C & NSTROP=1. IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.1)) GO TO 500 IF((FLOWIN.LE.0.0).AND.(FLOBOT.GE.0.0)) GO TO 500 C C19------IF HEAD > BOTTOM THEN ADD TERMS TO RHS AND HCOF. C*** reset top active layer if dry cell IF(IBOUND(IC,IR,IL).EQ.0) IL=NIL IF(IQFLG.GT.0) GO TO 400
Subroutine STR1BD This subroutine calculates volumetric budget for streams. SUBROUTINE STR1BD(NSTREM,STRM,ISTRM,IBOUND,MXSTRM,HNEW,NCOL,NROW, 1 NLAY,DELT,VBVL,VBNM,MSUM,KSTP,KPER,ISTCB1,ISTCB2,ICBCFL,BUFF, 2 IOUT,NTRIB,NSS,ARTRIB,ITRBAR,IDIVAR,NDFGAR,ICALC,CONST,IPTFLG, 3 NSTROP) ... skip lines... C14----DETERMINE LEAKAGE THROUGH STREAMBED. IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.1)) GO TO 315 IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.3)) THEN DO 311 NIL=IL+1,NLAY C C18A-----IF CELL IS: CONSTANT HEAD MOVE ON TO NEXT STREAM REACH C INACTIVE MOVE DOWN A LAYER C ACTIVE SET LEAKAGE TO CONSTANT IF(IBOUND(IC,IR,NIL)) 3111,311,3112 311 CONTINUE C------NO ACTIVE CELLS FOUND, NO LEAKAGE 3111 FLOBOT=0. GO TO 320 3112 IF (STRM(3,L).ne.0) & WRITE (IOUT,900) NIL, ISTRM(4,L), ISTRM(5,L), IR, IC 900 FORMAT (/,5X,'LEAKAGE TO LAYER',I5,' FROM STREAM SEG', & I6,' STREAM REACH',I6,' IN ROW',I5,' COLUMN',I5) ENDIF IF(FLOWIN.LE.0.0) HSTR=STRM(5,L) CSTR=STRM(3,L) SBOT=STRM(4,L) H=HNEW(IC,IR,IL) C*** set head to top active layer if dry cell IF(IBOUND(IC,IR,IL).EQ.0) H=HNEW(IC,IR,NIL) T=HSTR-SBOT ... skip lines... C18----STREAMFLOW OUT EQUALS STREAMFLOW IN MINUS LEAKAGE. 315 IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.1)) FLOBOT=0. 320 FLOWOT=FLOWIN-FLOBOT IF((ISTSG.GT.1).AND.(NREACH.EQ.1)) STRM(9,LL)=ARTRIB(IFLG) C C19----STORE STREAM INFLOW, OUTFLOW AND LEAKAGE FOR EACH REACH. STRM(9,L)=FLOWOT
66 STRM(10,L)=FLOWIN STRM(11,L)=FLOBOT C C20----IF LEAKAGE FROM STREAMS IS TO BE SAVED THEN ADD RATE TO BUFFER. C------OPTION=3; LEAKAGE IS INTO HIGHEST CELL IN A VERTICAL COLUMN C------THAT IS NOT NO FLOW. IF NO ACTIVE CELLS EXIST THEN ZERO STREAM C------LEAKAGE IF ((IBD.EQ.1).AND.(FLOBOT.NE.0.0)) THEN IF (IBOUND(IC,IR,IL).GT.0) BUFF(IC,IR,IL)=BUFF(IC,IR,IL)+FLOBOT ELSE BUFF(IC,IR,NIL)=BUFF(IC,IR,NIL)+FLOBOT ENDIF ENDIF C C21----DETERMINE IF FLOW IS INTO OR OUT OF MODEL CELL. ... skip lines... C29-----SAVE STREAMFLOWS OUT OF EACH REACH ON DISK. C DO 615 L=1,NSTREM IC=ISTRM(3,L) IR=ISTRM(2,L) IL=ISTRM(1,L) IF((IBOUND(IC,IR,IL).LE.0).AND.(NSTROP.EQ.1)) GO TO 615 IF (IBOUND(IC,IR,IL).GT.0) & BUFF(IC,IR,IL)=BUFF(IC,IR,IL)+STRM(9,L) C*** set layer to top active cell IF (IBOUND(IC,IR,IL).EQ.0.AND.NSTROP.EQ.3) & BUFF(IC,IR,NIL)=BUFF(IC,IR,NIL)+STRM(9,L) 615 CONTINUE
67 APPENDIX 3: MULTI-AQUIFER WELL PACKAGE SUBROUTINE MAW5AL(ISUM,LENX,MXMAW,NMAWS,IN,IOUT, 1 NLAY,LCRMAW,LCRMAL,IMAWCB) C C-----VERSION 5.0 OCT2001 MAW5AL C ****************************************************************** C ALLOCATE ARRAY STORAGE FOR MULTI-AQUIFER WELL PACKAGE C ****************************************************************** C C1------IDENTIFY PACKAGE AND INITIALIZE NMAWS WRITE(IOUT,1) 1 FORMAT(1H0,’MAW5 -- MULTI-AQUIFER WELL PACKAGE VERSION 5’) NMAWS=0 C C2------READ MAX NUMBER OF MAWS AND UNIT OR FLAG FOR C2------CELL-BY-CELL FLOW TERMS. READ(IN,2) MXMAW,IMAWCB 2 FORMAT(2I10) WRITE(IOUT,3) MXMAW 3 FORMAT(1H ,’MAXIMUM OF’,I5,’ MULTI-AQUIFER WELL CATEGORIES’) IF(IMAWCB.GT.0) WRITE(IOUT,9) IMAWCB 9 FORMAT(1X,’CELL-BY-CELL FLOW WILL BE SAVED ON UNIT’,I3) IF(IMAWCB.LT.0) WRITE(IOUT,8) 8 FORMAT(1X,’CELL-BY-CELL FLOWS WILL BE PRINTED WHEN ICBCFL NOT 0’) C C3------ALLOCATE SPACE FOR ARRAYS RMAW AND RMAL. ISOLD=ISUM LCRMAW=ISUM ISUM=ISUM+5*MXMAW LCRMAL=ISUM ISUM=ISUM+4*NLAY*MXMAW ISP=ISUM-ISOLD C C4------PRINT NUMBER OF WORDS IN X ARRAY USED BY MAW PACKAGE. WRITE(IOUT,4) ISP 4 FORMAT(1X,I6,’ ELEMENTS IN X ARRAY ARE USED FOR MAWS’) ISUM1=ISUM-1 WRITE(IOUT,5) ISUM1,LENX 5 FORMAT(1X,I6,’ ELEMENTS OF X ARRAY USED OUT OF ’,I13) IF(ISUM1.GT.LENX) WRITE(IOUT,6) 6 FORMAT(1X,’ ***X ARRAY MUST BE DIMENSIONED LARGER***’) C C5------RETURN RETURN END C
68 SUBROUTINE MAW5RP(RMAW,RMAL,NMAWS,MXMAW,IN,IOUT, 1 NLAY) C-----VERSION 5.0 OCT2001 MAW1RP C ****************************************************************** C READ MULTI-AQUIFER WELL LOCATIONS AND STRESS RATES C modified by RMY to omit categories & read data from single line C ****************************************************************** C C SPECIFICATIONS: C ------DIMENSION RMAW(5,MXMAW),RMAL(4,NLAY,MXMAW) INTEGER LAYERS(10) C ------C C1------READ ITMP(# OF MAW WELLS OR FLAG SAYING REUSE MAW DATA) READ (IN,1) ITMP 1 FORMAT(I10) IF(ITMP.GE.0) GO TO 50 C C2------IF ITMP LESS THAN ZERO REUSE DATA. PRINT MESSAGE AND RETURN. WRITE(IOUT,6) 6 FORMAT(1H0,’REUSING MULTI-AQUIFER WELLS FROM LAST STRESS PERIOD’) GO TO 260 C C3------ITMP=>0. SET NMAWS EQUAL TO ITMP. 50 NMAWS=ITMP IF(NMAWS.LE.MXMAW) GO TO 100 C C4------NMAWS > MXMAW. PRINT MESSAGE. STOP. WRITE(IOUT,99) NMAWS,MXMAW 99 FORMAT(1H0,’ITMP(’,I4,’) IS GREATER THAN MXMAW(’,I4,’)’) STOP C C5------PRINT NUMBER OF MAW WELLS IN CURRENT STRESS PERIOD. 100 WRITE (IOUT,2) NMAWS 2 FORMAT(1H0,/ + 1X,I5,’ MULTI-AQUIFER WELLS’) C C6------IF THERE ARE NO ACTIVE MAWS IN THIS STRESS PERIOD THEN RETURN IF(NMAWS.EQ.0) GO TO 260 C C7------PRINT HEADING FOR MAW INPUT. WRITE(IOUT,3) 3 FORMAT(1X,/ + 1X,’ ROW COL M.A.W. RATE RADIUS RATIO WELL NO.’,/ + 1X,’------’) DO 250 II=1,NMAWS C C8------FOR EACH WELL READ AND PRINT ROW, COLUMN, RATE, C8------# OF LAYERS SCREENED AND # IN CATEGORY. READ (IN,4) I,J,Q,RRATIO,NUM,(LAYERS(L),L=1,10) 4 FORMAT(2I10,2F10.0,I5,10I3) WRITE (IOUT,7) I,J,Q,RRATIO,II 7 FORMAT(2X,I8,I7,G16.5,F10.2,I8) RMAW(1,II)=I RMAW(2,II)=J RMAW(4,II)=RRATIO C*** Q in Bennett et al (1982) is positive to for discharging well C*** Q in this code is read as negative for discharging well RMAW(5,II)=-Q NWLAYS=0 DO 230 L=1,10 IF (LAYERS(L).EQ.0) GO TO 230 NWLAYS=NWLAYS+1 RMAL(1,NWLAYS,II)= L
69 230 CONTINUE IF ((NWLAYS.ne.num).AND.(Q.ne.0)) THEN WRITE (IOUT,9) NUM,II 9 FORMAT(5x,’NUMBER OF LAYERS NOT EQUAL’,I5,’ FOR WELL #’,I5) STOP ENDIF RMAW(3,II)=NWLAYS WRITE(IOUT,8) (RMAL(1,JJ,II), JJ= 1,NWLAYS) 8 FORMAT(1X,’LAYERS: ’,10f5.0) DO 240 JJ=1,NWLAYS C C9------FOR EACH SCREENED LAYER READ LAYER # AND RADIUS RATIO. RMAL(2,JJ,II)=RRATIO 240 CONTINUE 250 CONTINUE C C10-----RETURN 260 RETURN END C SUBROUTINE MAW5FM(NMAWS,MXMAW,RHS,HCOF,IBOUND,RMAW, 1 RMAL,NCOL,NROW,NLAY,CR,DELC,DELR,HNEW,IOUT) C C-----VERSION 5.0 OCT2001 MAW5FM C C ****************************************************************** C ADD MULTI-AQUIFER WELL FLOW TO RHS AND HCOF C ****************************************************************** C C SPECIFICATIONS: C ------DOUBLE PRECISION HNEW C DIMENSION RHS(NCOL,NROW,NLAY),HCOF(NCOL,NROW,NLAY), 1 RMAW(5,MXMAW), 2 RMAL(4,NLAY,MXMAW),CR(NCOL,NROW,NLAY),DELC(NROW), 3 DELR(NCOL),HNEW(NCOL,NROW,NLAY),IBOUND(NCOL,NROW,NLAY) C ------PI=3.14159 C C1------PROCESS EACH MAW CATEGORY DO 90 I1=1,NMAWS C C2------GET THE INFORMATION FOR THE CATEGORY NWLAYS=RMAW(3,I1) I=RMAW(1,I1) J=RMAW(2,I1) SUMF=0 SUMFH=0 C C3------PROCESS THE CELL IN EACH SCREENED LAYER DO 30 I2=1,NWLAYS K=RMAL(1,I2,I1) C C4------IF THE CELL IS NOT ACTIVE MOVE ON TO THE NEXT LAYER. IF(IBOUND(J,I,K).LE.0) GO TO 30 C C5------CALCULATE THE TRANSMISSIVITY OF THE CELL. TRM=CR(J-1,I,K)*((DELR(J-1)+DELR(J))/2)/DELC(I) TRP=CR(J,I,K)*((DELR(J)+DELR(J+1))/2)/DELC(I) IF ((TRM.NE.0).AND.(TRP.NE.0)) GO TO 50 C C6----IF EITHER TRANSMISSIVITY IS ZERO THEN CHECK TO SEE WHETHER C WELL EXTENDS TO LOWER LAYERS, IF NOT, STOP SIMULATION IF (I2.LT.NWLAYS) GO TO 30
70 WRITE(IOUT,888) K,I,J 888 FORMAT(1H ,’MAW FAILS. WELL IN LAYER’,I5,’ ROW’,I5,’ COLUMN’, 1 I5,’ IS ADJACENT TO AN INACTIVE CELL’) WRITE(IOUT,889) 889 FORMAT(1H ,’SIMULATION ENDING’) STOP C C7------CALCULATE F AND FH, STORE THEM IN ARRAY RMAL AND ADD THEM C7------TO ACCUMULATORS SUMF AND SUMFH. 50 TR=2*TRP*TRM/(TRP+TRM) RRATIO=RMAW(4,I1) F=TR/(ALOG(RRATIO)) SUMF=SUMF+F RMAL(3,I2,I1)=F HTMP=HNEW(J,I,K) FH=F*HTMP SUMFH=SUMFH+FH RMAL(4,I2,I1)=FH 30 CONTINUE C*** check for dry wells IF (sumf.eq.0) THEN WRITE(IOUT,890) I, J 890 FORMAT(’ DRY WELL IN ROW ’,I5, ’ COLUMN ’,I5) GO TO 90 ENDIF C C8------CALCULATE THE HEADS IN THE WELLS IN THIS CATEGORY Q=RMAW(5,I1) HWELL=(SUMFH/SUMF)-(Q/(2*PI*SUMF)) C C9------FOR EACH CELL ADD TERMS FOR THIS CATEGORY TO HCOF AND RHS. C*** categories omitted in this version DO 60 I2=1,NWLAYS K=RMAL(1,I2,I1) IF(IBOUND(J,I,K).LE.0) GO TO 60 F=RMAL(3,I2,I1) NINCAT=1 HCOF(J,I,K)=HCOF(J,I,K)-2*PI*F*NINCAT RHS(J,I,K)=RHS(J,I,K)-2*PI*F*HWELL*NINCAT 60 CONTINUE 90 CONTINUE C C10-----RETURN RETURN END SUBROUTINE MAW5BD(NMAWS,MXMAW,IBOUND,RMAW, 1 RMAL,NCOL,NROW,NLAY,HNEW,KSTP,KPER,IMAWCB, 2 ICBCFL,BUFF,IOUT,MSUM,DELT,VBNM,VBVL) C-----VERSION 5.0 OCT2001 MAW5BD C C ****************************************************************** C CALCULATE CELL-BY-CELL FLOWS AND BUDGET TERMS FOR C MULTI-AQUIFER WELLS C ****************************************************************** C C SPECIFICATIONS: C ------DOUBLE PRECISION HNEW C DIMENSION HNEW(NCOL,NROW,NLAY),IBOUND(NCOL,NROW,NLAY), 1 RMAW(5,MXMAW), 2 RMAL(4,NLAY,MXMAW), 3 BUFF(NCOL,NROW,NLAY),VBVL(4,20),VBNM(4,20) DIMENSION TEXT(4) CHARACTER*4 TEXT,VBNM
71 DATA TEXT(1),TEXT(2),TEXT(3),TEXT(4) /’MULT’,’I-AQ’,’IFR ’,’WELL’/ C ------C C1------CLEAR RATIN AND RATOUT ACCUMULATORS. IBD=0 RATIN=0. RATOUT=0. C C2------IF THERE ARE NO MAWS DO NOT ACCUMULATE FLOW IF(NMAWS.EQ.0)GO TO 200 C C3------TEST TO SEE IF CELL-BY-CELL FLOW TERMS WILL BE RECORDED. IF(ICBCFL.EQ.0 .OR. IMAWCB.LE.0 ) GO TO 10 C C4------IF CELL-BY-CELL FLOWS WILL BE SAVED THEN CLEAR THE BUFFER. IBD=1 DO 5 IL=1,NLAY DO 5 IR=1,NROW DO 5 IC=1,NCOL BUFF(IC,IR,IL)=0. 5 CONTINUE C C5------PROCESS MAW CATEGORIES ONE AT A TIME. 10 PI=3.14159 DO 150 I1=1, NMAWS NWLAYS=RMAW(3,I1) I=RMAW(1,I1) J=RMAW(2,I1) SUMF=0 SUMFH=0 C C5A-----FOR EACH CELL OPEN TO THE CATEGORY CALCULATE F AND FH DO 30 I2=1,NWLAYS K=RMAL(1,I2,I1) IF(IBOUND(J,I,K).LE.0) GO TO 30 F=RMAL(3,I2,I1) SUMF=SUMF+F HTMP=HNEW(J,I,K) FH=F*HTMP SUMFH=SUMFH+FH RMAL(4,I2,I1)=FH 30 CONTINUE C*** skip over if dry well IF (sumf.eq.0) GO TO 150 C C5B-----CALCULATE THE HEAD IN THE WELLS IN THIS CATEGORY Q=RMAW(5,I1) HWELL=(SUMFH/SUMF)-(Q/(2*PI*SUMF)) C C5C-----FOR EACH LAYER IN WHICH THE MAW IS SCREENED PROCESS C5C-----THE CELL WHICH CONTAINS THE MAW. DO 100 I2=1,NWLAYS C C5C1----CALCULATE RATE OF FLOW FROM THE MAWS INTO THE CELL. K=RMAL(1,I2,I1) IF(IBOUND(J,I,K).LE.0) GO TO 100 F=RMAL(3,I2,I1) HTMP=HNEW(J,I,K) NINCAT=1 RATE=2*PI*F*(HWELL-HTMP)*NINCAT C C5C2----IF BUDGET TERMS ARE TO BE SAVED THEN ADD RATE TO BUFFER. IF(IBD.EQ.1) BUFF(J,I,K)=BUFF(J,I,K)+RATE C C5C3----PRINT THE INDIVIDUAL RATES IF REQUESTED(IMAWCB<0).
72 IF(IMAWCB.LT.0.AND.ICBCFL.NE.0) WRITE(IOUT,900) (TEXT(N),N=1,4), 1 KPER,KSTP,I1,K,I,J,RATE,HWELL 900 FORMAT(1H0,4A4,’ PERIOD’,I3,’ STEP’,I3,’ MAW’,I4, 1 ’ LAYER’,I3,’ ROW ’,I4,’ COL’,I4,’ RATE’,G15.7, 2 ’ HWELL’,G15.7) IF(RATE) 90,100,80 C C5C4---- RATE IS POSITIVE(RECHARGE). ADD IT TO RATIN. 80 RATIN=RATIN+RATE GO TO 100 C C5C5----RATE IS NEGATIVE(DISCHARGE). ADD IT TO RATOUT. 90 RATOUT=RATOUT-RATE 100 CONTINUE 150 CONTINUE C C6------IF CELL-BY-CELL TERMS WILL BE SAVED THEN CALL UBUDSV TO C6-----RECORD THEM ON DISK IF(IBD.EQ.1) CALL UBUDSV(KSTP,KPER,TEXT,IMAWCB,BUFF,NCOL,NROW, 1 NLAY,IOUT) C C7------MOVE RATES INTO VBVL FOR PRINTING BY MODULE BAS1OT. 200 VBVL(3,MSUM)=RATIN VBVL(4,MSUM)=RATOUT C C8------MOVE RATES TIMES TIME STEP LENGTH INTO VBVL ACCUMULATORS. VBVL(1,MSUM)=VBVL(1,MSUM)+RATIN*DELT VBVL(2,MSUM)=VBVL(2,MSUM)+RATOUT*DELT C C9------MOVE BUDGET TERM LABELS INTO VBNM FOR PRINTING. VBNM(1,MSUM)=TEXT(1) VBNM(2,MSUM)=TEXT(2) VBNM(3,MSUM)=TEXT(3) VBNM(4,MSUM)=TEXT(4) C C10-----INCREMENT BUDGET TERM COUNTER(MSUM). MSUM=MSUM+1 C C11-----RETURN RETURN END
73 C.E. Heywood and R.M. Yager—SIMULATED GROUND-WATER FLOW IN THE HUECO BOLSON, AN ALLUVIAL-BASIN AQUIFER SYSTEM NEAR EL PASO, TEXAS—U.S. Geological Survey Water-Resources Investigations Report 02-4108 U.S. Department of the Interior WRD U.S. Geological Survey, NE, Suite 400 5338 Montgomery Blvd. NM 87109-1311 Albuquerque, BOOK RATE